-- Posted by EdZ on 10:51 am on April 6, 2003 Just to review past posts: Joystick(s) - advantages simple, easy to configure, intuitive, usually all that is needed for a task. disadvantages - hard to perform complex actions (arm manipulation, leg specific motion, etc), buttons can be confusing and inadequate for certain functions VR suit: advantages - can move whole body of mech intuitively, little prior training needed, accurate manipulation disadvantages - hard to aim weapon (no direct crosshair movement), can harm pilot if mech destabilises (unles in a needlesly heavy and expensive, not to say bulky, gyroscopically stabilised cockpit), very cpu intensive and hard to program control system, and may require VR helmet (very expensive), no buttons to control simple functions. Now on to my idea. Many people when talking about the joystick method of control refer to Gundam wing. What many people don't notice is that although the pilots hold joysticks (well, rods with buttons on anyway), most of the actual control is through movement of the arms of the chair that they sit in (which the joysticks are attached to the end of). this is not useful in itself, unless the idea is developed a bit. The ability to easily control a joystick can be combined with the dexterity of a VR suit to create a hybrid control system with the advantages of both approaches. The pilot can hold the joysticks in his hands, and have each mayor body part (upper/lower arm/leg) strapped to a semi-mobile section of his chair. The torso of the pilot hould be held against the back of the chair by a secure harness, and his legs would be free to move beneath him, whilst still giving support. The moving sections could be locked into chair position for simple operations, and could be unlocked to move freely when more precise motion is required. the attachment of the pilot to the chair prevents him 'bouncing about' if the mech were to topple and lower the risk of injury. the joints would be simple to program, as they could only move in a fixed area, and would be continually connected, and no moving sections of cockpit would be required (eg, the treadmill idea). -- Posted by Ajay on 12:50 pm on April 6, 2003 <Joystick(s) - advantages simple, easy to configure, intuitive, usually all that is needed for a task. disadvantages - hard to perform complex actions (arm manipulation, leg specific motion, etc), buttons can be confusing and inadequate for certain functions>>   The joysticks could have XYZ mpovement, meaning they could moved up, down, left, right, AND back and forth.  That opens up a wider range of motion, especially for the arms.  That would be very useful for a wide variety of tasks, although the manipulation of individual digits can be of some concern.   The legs controls for such a sytem could be foot operated throttles, or preferably hatswitch style controls.  you have 2 sliding panels your feet lock into.  Push both forward, vehicle goes forward, pull both back, vehicle goes back, push one forward and one back, and the vehicle goes back.  Shift or tilt to the right or left in order to strafe or side step. <disadvantages - hard to aim weapon (no direct crosshair movement), can harm pilot if mech destabilises (unles in a needlesly heavy and expensive, not to say bulky, gyroscopically stabilised cockpit), very cpu intensive and hard to program control system, and may require VR helmet (very expensive), no buttons to control simple functions. >>   Balance also becomes a bit of a problem, especially if the vehicle isn't built exactly to human proportions...... <>   The joystick/hatswitch design is actually more inspired by Gasaraki, Patlabor, or Heavy gear.   However, there have been many suggestions about a conventional joystick method combined with motion capture, but lower body motion capture, in my opinion, is not a very effective way of operating a vehicle's movement.  It's best left to the arms and upper torso if anything. (Edited by Ajay at 11:53 am on April 6, 2003) (Edited by Ajay at 3:30 pm on April 7, 2003) -- Posted by EdZ on 3:13 pm on April 6, 2003 Like the sliding foot throttle idea! All the improvements in jowstick design etc could be implemented on this design. The only extra would be the ability to disengage the chair and do more complex actions requiring more dexterity than can be provided by the conventional controls. if i have any spare tim in the week ill try to upload a pic of what i mean, to make it a bit clearer. -- Posted by Caesar02 on 4:25 pm on April 6, 2003 Ed..Z... EdZ, I really like your idea. It might take a little getting used to for the pilot, but I think it might work. Of course, I still like my idea better: using a PS2 or Xbox controller. -- Posted by EdZ on 4:42 pm on April 6, 2003 I beat you to the punch on that one. I play Mechwarrior on my pc using a PS2 controller connected up through an EMS USB2 adaptor. Ha! -- Posted by ALostEE on 6:14 pm on April 6, 2003 Three Words try a Nintendo Power glove. well you may need 2 to get full control but it think it will work perfectly for the application. -- Posted by JB on 1:35 am on April 7, 2003 Quote: from EdZ on 9:51 am on April 6, 2003 Many people when talking about the joystick method of control refer to Gundam wing. What many people don't notice is that although the pilots hold joysticks (well, rods with buttons on anyway), most of the actual control is through movement of the arms of the chair that they sit in (which the joysticks are attached to the end of). Actually, in my experience what they don't notice is that it's a work of fiction and often the same control motions produce radically diffrent effects. -- Posted by EdZ on 2:54 am on April 7, 2003 I was just trying to make an example! -- Posted by Manae on 10:15 pm on April 8, 2003 Okay, now this is something I've genuinely thought long and hard on...  Controling a mech would be one of the most difficult things in the world. **Most of what I'm saying is based on military design, though it can also make sense otherwise** I wrote up a rough article on it here:  http://yamayo.dragonwings.org/~shard/mechcheck/  under the 'essay' section.  It's incomplete and rushed, but has the basic ideas (though, I sent in an update that's not been added yet.  The site's...  not updated much:P) In anycase, I've come to a pretty simple conclusion:  most complex actions, i.e. aiming, switching weapons in the case of manipulators, would be programmed actions.  The pilot would flip a few buttons (say:  a routine access, what routine to perform, and an execute) and the mech would do it, hopefully un-interrupted. Oh, as for joysticks, don't forget that it's not just XYZ, but you can also toss in a twisting motion and moving the entire joystick (like in Gundams, and others maybe?).  It would be hard to pick up on, but adds more movement options. Like I said, I'm yet to really get a good idea on legs, but I like the sliding pedals.  It's better than simple back-and-forth by adding side-to-side, does allow for strafing...  But still lacks a good answer to rough terrain.  Differing magnitudes of push can change lenth of stride, but you still don't have enough control for, say, crouching.  It would take a complex system to have such manuevers programmed in, but maybe such things could be buttons to push on the console. Pick away as you please:) -- Posted by Earle Bishop on 12:05 am on April 9, 2003 http://mechaps.com/CMI/Pierre_Bergeron/ -- Posted by Ajay on 2:22 pm on April 9, 2003 <>  I was thiniking that the detailed processes of walking (leg and foot joint movement) would be taken care of by the vehicle itself, even in rough terrain. It'd be more as if the pilot were "riding" the vehicle.    For crouching, possibly a switch or dial that commands the vehicle to raise its torso higher or lower, with it taking appropriate stances and walking gaits depending on height.   Or maybe by using the joysticks to maneuver the arms down low enough, you could reach a point where the vehicle's arms can't continue moving in their fixed position, and the whole torso or body moves.  I was thinking of something like that for twisting the torso. -- Posted by Manae on 6:07 pm on April 9, 2003 Okay, that word document does give a good idea on how you could do it, though I find some points a little...  well, non-functional, for lack of a better description.  While I realize this isn't exactly the officially settled setup (or is it now?), here we are: Voice commands -- I don't think this is the best idea.  It has merit, but voices change.  Not only from person to person, but stress can alter patterns enough to throw off these systems.  It'd be a pain to calibrate even a few commands to a new pilot as well.  It's more of a user-friendly problem than an effectiveness one in the end.  Simply pushing a button would be a better idea, don't you think? Trace gloves -- *so it's a term from G Gundam, but it fits well to the function* Another good idea, but flawed.  What happens if the pilot has in itchy nose while on the job?  It'd be funny to see a mech scratching an imaginary nose, but you can't always have that.  What if you're holding a beam?  Okay, so turn off the trace, but if you can't have a voice command, you still can't do that.  You can't always move arms to turn off buttons, either.  Though again a possibility, there are bugs which need to be worked out first. As a note, I don't think you can claim the control ideas as you propriety, as I've seen all of those examples in use for a long time in anime, so technically while you are making the real thing, the ideas been around longer and you're copying off of it.  You can argue this veiw, but it can't be decided unless it had to be taken to court for some reason. Oh, and again knowing it isn't the actual material of MPS, I need to poke some fun at the mech pictures in that doc.  Despite the constant assertions that this is by no means a mobile suit, the first thing I thought of when I saw the mech in that document as a GM. :P -- Posted by Caesar02 on 9:59 pm on April 9, 2003 All I have to say is that I hope a professional mech operator has the self control to keep from scratching his nose while on the job. Besides, if you have voice command, you can just tell the mech to hold it's position until your done scratching. And also, voice command technology has come a long way. And its gonna keep on going. We dont have to worry about technology not being exactly where it needs to be right now, cuz the mech isnt rolling out of the factory tomorrow. By the time it is finished, either voice command will be a lot more effective or there will be a new and even better way to do it. Like neural links or something. You knever no. -- Posted by Ajay on 10:47 pm on April 9, 2003 <>   I always wondered what jet pilots do if they have an itchy nose, but can't take off their mask for one reason or another.  Or what about those guys who handle dangerous materials or microchips in those big bulky suits?  What about if an astronaut gets an itch somewhere while he's out working on a satelite?   besides, I think the motion capture is optional, and would probably be useful mainly for fine manipulation, when little else could trully substitute.   Voice activated controls have been developed for the Eurofighter jet, and have over 95% accuracy (about 95 or 98% I think).  Even then, they're not connected to any really critical systems (like operating the radio and displays) and there's still the option to activate something manually via a switch or button. <>   Reminded me of an Ingram from Patlabor :) -- Posted by Earle Bishop on 11:12 am on April 10, 2003 <> Not official, but a good representation of what we are shooting for. <> Voice recognition is a proven technology. Sometimes, but not always, a button is just as good or better. Complete voice control is not likely. Were it so, completely voice controlled operating systems like OS/2 would have taken off years ago. <> It is called "Motion capture" or simply "mocap." Are you an engineer? I love how you belayed the solution to create a problem. There is no reason to NOT use voice control. In fact, it resolves the problem of going into and out of mocap modes without having to change positions.   Pilot: Just a sec....   Mark1: Mocap Stanting by   (Pilot scratches nose)   Pilot: Ok, let's go   Mark1: Mocap online <> Considering BanDai has zero patents, etc, on industrial control devices that have been around far longer than gundam, I would say they are in a poor position to dispute. It's not like we sat down and watched an episode of GW or Gasaraki and drew up a neat interface. It took a lot of time and research to arrive at the conclusions we did. <> I swear, you are an engineer, aren't you. <> Actually, I think it is a gundam model. PL had a friend help him complete the animations. We ran short on time for a convention, and he had that model already, so they used it. Trust me, I complained  =) -- Posted by Manae on 3:06 pm on April 10, 2003 I'm not an engineer yet...  again, I can't tell if you're being sarcastic or not.  It helps to actually talk to people in situations like this.  Then again, since second grade I've been told I'm going to be an engineer even if it wasn't my first career plan:P One problem you can really see with voice commands is accidentally using it. *In a war, during a battle* Commander *intercom*:  Hurry up and do this! Pilot:  Just a sec, Sir, I can-- *MOCAP Disabled* Pilot: ... Of course, use that command to disable it, and you should get your head examined. -- Posted by Ajay on 3:47 pm on April 10, 2003  Mo-cap control has its advantages when it comes to finely manipulating random objects,such as handling hazardous materials and other jobs that requires human like dexterity.       Possible military applications include aiding in asembling structures, clearing up battlefield obstructions, and moving objects (including cargo and supplies).  These kinds of operations and tasks usually take up more time than the actual fighting.  Preferably, the switch over from mo-cap to conventional joystick control should be almost instantaneous, so the time taken to switch wouldn't really be putting you in mortal danger compared to any other system.   Why would the pilot be using mo-cap or fidling with systems in the midst of an intense fire fight anyway?      Actually, when the pilot's hands and eyes are busy is the perfect time to have something like voice actiavted controls to operate some of the less vital systems. (Edited by Ajay at 2:50 pm on April 10, 2003) (Edited by Ajay at 8:49 pm on April 11, 2003) -- Posted by Kusanagi on 7:23 pm on April 11, 2003 Hehe. the mech in the animation is indeed a Gundam.. heck.. I know its a poor choice, but we were going to host a panel at an Anime Convention and we were running too short on time to create our own model.  So yeah..  it is crap, but at least it gives a good idea of how the control interface works. ;) I think i specified in my technical document that the Motion Caption could be turned off via voice commands in order to use 2 joysticks fully calibrated to the pilot's liking.  So the Mo-Cap would only be used for special work such as construction work.   If mechs ever become military weapons, while in battle, they could use joysticks with an aiming system similar to the Longbow helicopter, which aims where the gunner looks. All the pilot would have to do is look at his target and press the trigger. :) Also the voice commands would never become an issue while talking via radio with other mech pilots.  The radio will be activated with a button near the hand (probably on the joystick or somewhere near the palm on the pilot's glove).  Its hard to explain but it is easily reachable with one finger and would not be easy to accidently activate.  While the pilot would be pressing the button to activate the radio, voice commands would be disabled.   -- Posted by JB on 4:31 am on April 12, 2003 Quote: from Earle Bishop on 10:12 am on April 10, 2003 <> It is called "Motion capture" or simply "mocap." Are you an engineer? I love how you belayed the solution to create a problem. There is no reason to NOT use voice control. In fact, it resolves the problem of going into and out of mocap modes without having to change positions.   Pilot: Just a sec....   Mark1: Mocap Stanting by   (Pilot scratches nose)   Pilot: Ok, let's go   Mark1: Mocap online More direct(if a tad sillier) solution: "Computer, scratch my nose." *nose scratching apparatus descends from the ceiling, scratches the nose, then retracts* ... That's dangerously close to Inspector Gadget there... -- Posted by SteelViper30 on 7:48 pm on May 4, 2003 Another good control option could be a neuro helmet. Something you put on that reads your brain waves so that you control the mech's movement with your subconscious (ex: leaning for balance) so that it's like just running around in a giant You, thus relieving you of operating the movement and focusiong on the mission at hand? -- Posted by Slayer85 on 8:42 pm on May 4, 2003 Quote: from SteelViper30 on 4:48 pm on May 4, 2003 Another good control option could be a neuro helmet. Something you put on that reads your brain waves so that you control the mech's movement with your subconscious (ex: leaning for balance) so that it's like just running around in a giant You, thus relieving you of operating the movement and focusiong on the mission at hand? Well, since they don't yet exist, other options are more practical for the time being. -- Posted by Fanboy on 5:22 pm on July 16, 2003       Why not have the first two fingers and the thumb slip through some rings attached to the joystick? It would provide some degree of motion capture and leave the other two fingers free for button mashing. Also, the mech could be set up heavy gear style, but with the pilot's arms in the mech's arms.  A regular dual-joystick setup would be fine if the mech was like, oh say, the Caldron Born (a 65 ton Heavy Battlemech). The neuro helmet is an exelent idea, but is way to advanced for it to be around in our lifetimes.   -- Posted by Earle Bishop on 7:00 pm on July 16, 2003 Why use a friggin joystick?  We're using motion capture hand waldos for manipulating the arms. -- Posted by EdZ on 2:09 am on July 17, 2003 This is what i was trying to get at in the first place! -- Posted by Caesar02 on 12:49 pm on July 17, 2003 I guess I'm not that good with this techie jargon, so... WTF is a waldo? -- Posted by ChrisDickerson on 2:12 pm on July 17, 2003 I work with a state of the art voice recognition system at the ware house. All it takes to throw it into an electronic tizzy is raising your voice. I play hell with the thing if I have a cold, or just an allergy attack. I'll admit, voice recognition has been around for a while... But I'll argue with anyone that says it's the end-all, be-all of controlling tech. I'd refuse to stake my life on it. The main problem comes when you're programming the intial words into the system. You're calm and unhurried. You voice inflection, which is what the system actually recognizes, is fairly flat. Get a cold, it gets' so falt, the system has a hard time understanding. Get angry or excited, and the inflection goes through the ceiling, and the system won't recognize it at all. It would take some special training to teach the pilot how to keep his voice inflections constant in order for a Voice Recognition system to be reliable. As far as hand controls go, you can rig joystick and throttle assemblies to have in excess of 20 controls that are within finger reach without taking your hands off the joystick or throttle. Hat-switches, several different kinds of push-buttons, 3-way toggles, pressure disks, knobs, you name it. And as far as Motion Capture goes, you could always activate and deactivate with a gesture that is not typical of people to reproduce without conciously thinking about it, like touching the fingernail of your pinky with the pad of your thumb. That requires thinking about it, and will prevent inadvertant switching, but is also something that can be done quickly without interrupting another task. -- Posted by Ghola on 4:37 pm on July 17, 2003 The waldo is a full body motion capture device that has seen use in a few movies and at least 1 anime convention. Movie.. Movie... have you seen FX? The movie FX, not the channel. At the end of it he controls a life-sized animetronic clown with a body suit. It is similar to that. We would not need the full body, but only the hands/arms, and maybe torso. I actually saw the Waldo when I first joined MPS at Anime Expo 2001. They were testing and calibrating it for the AMV contest. The system was hooked up to a computer with a 3D model of a girl, who was projected onto a massive screen. As the girl in the suit moved, the figure on the screen moved. There was a system for seeing when she opened her mouth, and she had a microphone so she could to the announcing. It was really neat. I havn't seen it in a while, but Earle knows more about it and the guy who made it. -- Posted by xyrix on 5:14 pm on July 17, 2003 maybe a system which combines a load of these principlies, if u had motion capture stuff for general movement, with computers correcting these movements and transating them into movements usable by the mecha, and perhaps cleaning up the movements, cos it would have to be highly precision, but be able to recogniceinvuluntary movements like trembling. (good in combat because of adrenaline shake, and good 4 caffiene junkies!!) thic couldbe combined with sensors which follow pupil movements for targeting, again with lots of computer involvement, perhaps if u moved your pupils over a number of targets while holding down a trigger, and to fire u could move back over then and press again. (would strain after a while...hmmm)of course if u are piloting a mission to fire missiles, say, at a specific target, a computer could handle that easaly leaving the pilot to dedicate his efforts to piloting. in fact, all ordanance coule be computer fired, and things such as close combat, where improvisation is needed could be left to the pilot. despatching a tank with a monofillament swird would be easy, but what about close combat between mechs?/ countermeasure systems mite be so effective that it ende up being the only way to destroy other mechs, then the battlefield woul;d change completly and be much closer to gundam style, close maching artillary and swords...omg ive gone off on one again!!! well its just a though n e way -- Posted by tomexe on 6:07 pm on July 17, 2003 Considering all that that Mk1 has is two arms and two legs a waldo for the hands is all that is really needed. More complex modes will probably require more complex controls. -- Posted by Fanboy on 7:22 pm on July 17, 2003  Sorry Earle, I must have read over the original Waldo mention.  Still, I have my doubts of motion capture being the primary control system for a combat mech. But that's probably my ignorance talking there. -- Posted by Earle Bishop on 1:47 am on July 18, 2003 We already have a joystick/throttle combo with a silly amount of options. Quick list: Masked fox fox A fox B flock indicator primary hat secondary hat tetrinary fire button (over forefinger reach, fox D) primary forefinger trigger secondary pinky trigger conventional joystick directions (dual axis) tetrinary hat control L/R sway rudder primary rotary switch (forefinger) secondary rotary switch (thumb) primary three phase mode switch secondary three phase mode switch mouse cursor control (4th hat) mouse click button fox D conventional throttle direction (single axis) I probably missed something... but you ge the jist of it. left and right hand P5 Waldos would be either voice control on/off, or one of the many finger accessible controls on the unit itself. They would only control the mech's hand movement. It all seems complex now, but as tomexe has stated, the future versions will put our very very simple systems to shame. -- Posted by Ajay on 2:20 am on July 18, 2003 << Sorry Earle, I must have read over the original Waldo mention.  Still, I have my doubts of motion capture being the primary control system for a combat mech. But that's probably my ignorance talking there. >>    The Mark-1 (Mechaps's first creation) is supposed to be a proof of concetp design.  The design itself is not supposed to display the exact potential abilities of a humanoid vehicle, just prove that it's possible to practically build one.  Wait for the US department of defense or some other more serious originization to get interested or involved to one degree or another before you see serious attempts at legged vehicles for military service (especially frontline combat).    So you're using a joystick/throttle system for operating both arms and legs?  What happened to the hatswitch style foot controls? I just kind of figured you'd use one joystick for each arm.   Also, do you mind explaining just what those functions on the throttle/control stick are supposed to make the Mark-1 istelf do?  Or is it totally user customized?   Has anyone ever done anything to control vehicles with little analog style control sticks like thouse found on most modern game console controllers? I was thinking that comething like that might be useful for placing a recticle or some kind of "targeter" on specific items or whatnot without moving the joysticks themselves or having to take your hands off. -- Posted by EdZ on 3:57 am on July 19, 2003 This is what i wasd trying to get at. 'traditional' controls (joysticks,foot throttles, etc) and mocop controls that can be switched in between easily. the main difference between the design i gave in the first post and the one MPS is using is that their design has a mocap 'suit' to capture the movements (at least thats what it sounds like from Earle's posts) and mine uses the chair the pilot sits in to capture the movements. Their desigh has an improved accuracy, and mine is (I think) easier to implement. -- Posted by Ajay on 12:54 pm on July 19, 2003   On the issue of a military humanoid vehicle.....   Since it must adhere to more rigid size constraints, it might be harder to use motion capture in any sense effectively.   Most military vehicles are limted to some 8-10 feet in width.  This not only helps them fit easily into smaller transports, but helps them pass between trees, get down narrow alleys, paths, etc.  As you'll often hear, one of the big annoying traits of many larger tanks is the fact that they're too large to go many places effectively.    If a motion capture requires that the pilot be able to fully extend his arms in either direction, you're pushing at least a 5-6 foot wide cockpit.  That leaves only 2-5 feet for additional stuff on the sides and the shoulder/weapons mounts.      I would imagine that a humanoid vehicle's cockpit would be relatively small and cramped, probably comparble to a small attack helicopter or small fighter (as to allow for smaller size) and would therefore limit the full potential of motion capture for even the arms.  Some are on the measure of only 3-4 feet wide.   Allthough this wouldn't be as much of a problem as with a much larger vehicle, a smaller humanoid would face this problem.  Any ways around this? -- Posted by tomexe on 12:15 am on July 20, 2003 I would imagine that a humanoid vehicle's cockpit would be relatively small and cramped, probably comparble to a small attack helicopter or small fighter (as to allow for smaller size) and would therefore limit the full potential of motion capture for even the arms.  Some are on the measure of only 3-4 feet wide.  Allthough this wouldn't be as much of a problem as with a much larger vehicle, a smaller humanoid would face this problem.  Any ways around this? There is the way they did it in Appleseed the landmates had armored "subarms" that projected out of the chest area that the operators own arms fit into. http://www.netsnapshot.com/pcw/buildalbum-cgi?ACCOUNT=1299&KEY=3 This is primarily useful for powerarmor type mechs The other is "point and click". Teach the mech to do the motions of a action by itself and you just tell it where and when. Only the most delicate tasks would require switching to motion tracking of the pilots hands, and for those jobs you would be stationary anyways. -- Posted by Caesar02 on 8:45 pm on July 23, 2003 You know, after the hundreds, nay, thousands (literally!) of mechs I have piloted, this is the conclusion I have reached: The Playstation 2 controller can be pretty damn effective. Hehehe -- Posted by noahrei on 9:54 pm on July 23, 2003 Is there much point to having to pass between the trees? They can always be knocked down you know. The way we are about conserving resources someone will be contracting for them anyways. -- Posted by tomexe on 10:50 pm on July 23, 2003 Quote: from noahrei on 6:54 pm on July 23, 2003 Is there much point to having to pass between the trees? They can always be knocked down you know. The way we are about conserving resources someone will be contracting for them anyways. Three reasons moving between trees is good. Increases speed Reduces trail Reduces wear and tear on mech. If you are in a situation where the above dont matter, like a whole mech army on the move with pleanty of repair vheicles around and the engineers having to come along anyways to take the trees down anyways to get the supply trucks in. Then sure, knock away, or use a ax. If you are trying to be stealthy, keep the mech concealed, not make noise and leave a clear trail of broken trees. If you are trying to make any good speed cross country and if you don't have any repair assets other than your own mech and your buddy mechs advancing with you. In these situations you want to avoid hitting trees with any force. -- Posted by ChrisDickerson on 5:46 pm on July 25, 2003 Besides, have you ever seen how big a Red Wood gets? Hard to imagine a mech knocking one of those out of the way. -- Posted by noahrei on 10:13 pm on July 25, 2003 At the risk of sounding patriotic, hopefully Americas mech force would be able to keep an enemy far enough away from the only redwoods on earth that he wouldn't have to wade in after him. But actually I am not sure what I was thinking then. It is very impractical but I beleiv trees have enough flexibility in most cases to allow a mech to pass. How was it done in Gundam 08th MS...or were the trees just always convieniently 40 feet apart. -- Posted by Caesar02 on 1:59 am on July 27, 2003 Actually, in old forests, the trees tend to be fairly far apart, and the ground below is almost completely clear. This is because the tall trees above almost completely block any sunlight that might reach the forest floor. If the trees are not this tall and do not prevent the growth of lesser plants below them, then that just means you have smaller trees all around you and you can cut yourself a path wherever you need to. I can tell you this much: I've got walking through the woods here in the foothills of north carolina. The trees around here are definetly dense enough that a mech couldn't just waltz through there walking around the vegetation. But this much is also true: it would be nearly impossible to spot a path cut through the woods from the ground. It would even be difficult to spot from the air. This is due mostly to the varying tree heights and the hillyness of the area. In short, it wouldn't be hard to hide a mech at my house. So, Earle, you can bring it over anytime after your finished. -- Posted by tomexe on 6:59 pm on July 27, 2003 Quote: from noahrei on 7:13 pm on July 25, 2003 At the risk of sounding patriotic, hopefully Americas mech force would be able to keep an enemy far enough away from the only redwoods on earth that he wouldn't have to wade in after him. But actually I am not sure what I was thinking then. It is very impractical but I beleiv trees have enough flexibility in most cases to allow a mech to pass. How was it done in Gundam 08th MS...or were the trees just always convieniently 40 feet apart. 08th MS team took place largely in South East Asia- Cambodia, Laos, parts of Vietnam and Thailand. They have triple canopy rainforest there that is more than tall enough for a good size mech to pass UNDER the topmost canopy. Scout helicopters in Vietnam used to fly underneath the layers, and the NVA used to make "tunnels" to shield their truck traffic by tying branches togeather. -- Posted by AP on 4:17 pm on Nov. 19, 2003 I've been thinking about this for awile and i think I have most of the problems solved. I propose that on armed mechs that you use two joysticks and two foot pedals. foot pedals contol the legs(tank like contols, one more forwardish then the oter to turn.)the joystics control the arms. triggers fire weapons and some simple voice capture to turn on and off simple stuff(power, controls, etc.) You would use the hat switch to aim or use the eye aiming like they have on sum helocopters. I've thought about this and all my colective knowledge of basic physics and engineering says it will work. (Edited by AP at 1:18 pm on Nov. 19, 2003) -- Posted by tomexe on 5:45 pm on Nov. 19, 2003 Cool, someone revived this old thread. Here are some drawings I posted on the concept database that show a hybrid waldo/point and click control chair from one of the Mortar Headds in Five Star Stories Cutaway: Exploded: In the world of Five Star Stories, the mecha spend a lot of time fencing with swords, in that case the pilot uses the waldo arm and hand controls. When NOT in combat, just navigating, the pilot is not required to take every step. Rather the pedals are used like those on a aircraft in conjunction with the hotas type control units under the hand area of the chairs arm units. The switch between the full, partial, and no motion tracking is on one of the HOTAS controllers. However, head movement and torso tiwst are always waldo, as they are controlled by the helmet and the shoulder braces, which while resembling the safety braces on a stand up rollercoster perform a dual role. They are a pressure sensetive control unit that generates commands for rotating and bending the torso but they are also a active device that acts to press the operator into their seat tightly to counter high G forces. The helmet support structure also performs a similar role holding the head and neck steady under G load, while also taking the weight of the helmet and its associated electronics off of the shoulders and neck of the pilot. (Edited by tomexe at 2:58 pm on Nov. 19, 2003) (Edited by tomexe at 3:05 pm on Nov. 19, 2003) -- Posted by kaempfer on 11:06 pm on Nov. 19, 2003 I have two good solutions to all arguments: The easy way: Use the controller from Steel Battalion The Old Skool Way: Make every mech function controllable with foot pedals and a steering wheel. -- Posted by Caesar02 on 11:54 pm on Nov. 19, 2003 <> The real way: use something practical, like foot pedals and waldos and voice command. -- Posted by EdZ on 3:24 am on Nov. 20, 2003 Not a good solution, but OK for testing purposes: Hok up a playstation controller and use armoured core/mechwarrioor controls. (mechwarrior works well with the dual analogue sticks). -- Posted by kaempfer on 10:50 pm on Nov. 20, 2003 Quote: from Caesar02 on 11:54 pm on Nov. 19, 2003 <> The real way: use something practical, like foot pedals and waldos and voice command. 'S got foot pedals.  Have you seen the controller?  There's like 64 buttons and 2 joysticks on the thing.  And for only 150 bucks! -- Posted by Caesar02 on 11:33 pm on Nov. 20, 2003 <<'S got foot pedals.  Have you seen the controller?  There's like 64 buttons and 2 joysticks on the thing.  And for only 150 bucks!>> Oh, I've seen it. But like I said, something PRACTICAL would work much better. Have you ever considered the fact that these things were designed as gaming tools, not actual vehicle controls? The idea of using junk like that for any real practical purpose is absurd. Yeah, so maybe the end result ends up using a couple joysticks and foot pedals. That doesn't validate the assesment that a video game controller is a practical way to control a vehicle. First of all, how are they even going to hook it up to the mech? And then, even if you get the whole interface thing to work, what are the chances that you can actually have a function for every button and toggle switch, or a toggle switch and button for every function? -- Posted by ChrisDickerson on 7:14 pm on Nov. 21, 2003 Ummm, the PCI is currently supposed to be a Saitek 45 joystick and throttle... That's a game controller set-up. Remember, right now, cheap is the word. Not in quality, but in price and availability. -- Posted by Viking420 on 6:48 pm on June 6, 2004 I was thinking of a system to add to the joystick pedal system. First off, a tank style pedal setup is no good for a mech, a tank can turn on a dime like a mech but a mech can also sidestep (ideally at least). Hatswitch pedals with an up/down axis added in, with rotation as well is my envisioned controll setup, but with an extra smaller pedal for the toes. At the very least the toe switch could be used to engage or disengage any articulated claws on the feet to grip weird terrain more securely. Dual joysticks are more than enough I think for upper body controls, use a trackball on top for shoulder controll, the joysticks themselves for elbow/wrist motions and either a mousewheel or button for "hand" controll. Full feedback (through a computer, not directly coupled) for all the controlls would help you keep your balance and guage whether you should keep pushing or not. -- Posted by kaempfer on 10:39 pm on June 6, 2004 Someone change his title to "necromancer". -- Posted by cygnus on 11:18 pm on June 25, 2004 if you basicly make a mech like the ones in a video game say mech assualt or steel battalion... then why wouldnt you just controll it like you do the game using a controller. its simple and inexspenive. for the concept of using a rotating upper torso like in a mad cat you could just use a joystick for leg movement and a joystick for upper body movement... say aiming a weapon while simaltaneously turning the cockpit like in mechassault. (Edited by cygnus at 8:23 pm on June 25, 2004) (Edited by cygnus at 8:27 pm on June 25, 2004) -- Posted by JB on 2:01 am on June 26, 2004 Quote: from cygnus on 10:18 pm on June 25, 2004 if you basicly make a mech like the ones in a video game say mech assualt or steel battalion... then why wouldnt you just controll it like you do the game using a controller. its simple and inexspenive. for the concept of using a rotating upper torso like in a mad cat you could just use a joystick for leg movement and a joystick for upper body movement... say aiming a weapon while simaltaneously turning the cockpit like in mechassault. Games are games. Reality is not games. While games have some good concepts in them, others work only within the realm of the game. -- Posted by cygnus on 5:46 pm on June 27, 2004 a game is still a basic simulation so why wouldnt it work on the same basic concepts -- Posted by EdZ on 5:59 pm on June 27, 2004 Because the mech is far from basic. -- Posted by cygnus on 11:14 pm on June 27, 2004 yes it is far from basic but it would still use those basic concepts.  what wouldn you be able to control with those same concepts besides things like cooling or power -- Posted by EdZ on 4:04 am on June 28, 2004 A-HAH! Cooling and power are only two of many, many things that must be controlled. Sure, the main controls can be simplified to an extreme degree, but the whole idea of the mech is for it to be versitile. IF it's range of motions are limited by simple controls then it can't do as many things. -- Posted by Earle Bishop on 4:11 am on June 28, 2004 No. I think you are underestimating the amount of systems involved. There are 72 different gauges in the cockpit, just for the backup system on the engines alone. Use this as an example; would you use a PS2 controller to drive a car? You MIGHT be able ot navigate around the streets, driving slowly, but it lacks simple components of how a car operates. Simply having a steering wheel allows you to quickly and accurately change directions. A direction pad does not. An accellerator or clutch allows you to establish a definitive "sweet spot" for desired reults. It is analog, connectred. It has infinite control levels. A game controller does not have this capability. A hamster wheel operates on the same basic concept as a car wheel. A hamster makes a poor choice for powering your car's wheel motion. -- Posted by Caesar02 on 9:19 am on June 28, 2004 <> Just to play devil's advocate here... The PS2 controller has analog input that can recognize different amounts of pressure. Also, video game controllers have been used to controlling cars for ages. Most people that play driving games enough can control the game car just as accurately as a real car, if not more so. The only reason that a steering wheel and pedals seems more useful to most people is that those are the controls that they have spent hundreds and hundreds of hours learning and using. -- Posted by EdZ on 10:01 am on June 28, 2004 Tha 'analogue' controls on a game controller are not truly anaolgue, they just have several 'steps' of movement. The Playstation controler has 9 'steps' for each axis on the analogue sticks. Cars have been made with sticks or pads to control them, but even after a lot of training people just couldn't drive them as well as with a steering wheel. -- Posted by Viking420 on 10:34 am on June 28, 2004 A few video game controllers might actualy work pretty well for non critical controlls. I wouldn't use them for navigation, but they might be a good cheap way to work out the upper body controlls. For the lower body movement and navigation you could use the chair itself as a controll, kind of like a motorcycle. Lean whatever way you want to go, the mech will "fall" in that direction and take a step that way to keep itself up (walking is essentialy perpetually falling forward). The foot pedals could then be simplified and freed for use controlling other aspects of the vehicle like general posture and gait. -- Posted by JB on 7:39 pm on June 28, 2004 Quote: from EdZ on 9:01 am on June 28, 2004 Tha 'analogue' controls on a game controller are not truly anaolgue, they just have several 'steps' of movement. The Playstation controler has 9 'steps' for each axis on the analogue sticks. 9 steps? I'm pretty sure you're mistaken. Virtually every pot-based input device, PS thumbsticks included, goes from 0-255. -- Posted by EdZ on 2:57 am on June 29, 2004 Nope. I opened it up, checked the potentiometer. Even plugged it into the PC adaptor and checked. I could only get 9 steps. It ma just have been a cheaply made controller, but I doubt it. -- Posted by geneofols on 10:28 am on June 29, 2004 I don't think it's a cheaply made controller.  I think it's a ceaply made adapter or a cheaply made driver on the PC.  I have my Xbox controller plugged into my PC, with a good driver, and it at least appears to be true analog, meaning it's either analog, or digital with the 0-255 range, with locations 127 and 128 being neutral. By the way, the range of movement applies to both the thumbsticks and the triggers on the Xbox gamepad, but the triggers' neutral position is 0. -- Posted by JB on 2:25 pm on June 29, 2004 Quote: from EdZ on 1:57 am on June 29, 2004 Nope. I opened it up, checked the potentiometer. Even plugged it into the PC adaptor and checked. I could only get 9 steps. It ma just have been a cheaply made controller, but I doubt it. It's a pot. There are no steps. They're determined by the reader. And for gaming purposes, a full byte dedicated to reading the pot gives them 256 steps, which is more steps than a human can readily detect. It also makes for faster processing, since most processors are optimized for byte reads.   The 256 levels on the DualShock 2 buttosn is the published spec. And I refuse to believe they read the buttons to a higher resolution than the pots. Also notable... http://www.gamesx.com/controldata/psxcont/psxcont.htm Comm protocol for a PS1. 1 byte per axis, range of 00-FF. Or 0-256, in decimal. In all analog input implementations. Even the mouse. I DO admit that having 256 positions is overkill for a thumbstick, though. Very few people can use them anywhere near that precisely. -- Posted by shadowhawk on 4:52 am on July 14, 2004 Well i'm new here and the way it looks every idea that has been inside has been said. Helment, suit, gloves, boots which can let the pilot walk and fly to and understand the weight and all. But you could keep the joysticks though. Basiclly like second command just incase something fails inside and the other contrals don't seem to respond. Like a backup. -- Posted by cakefool on 12:04 pm on July 14, 2004 I don't think anyone has a problem with a sensible simple backup joystick.  I sure would want one. -- Posted by cakefool on 11:05 am on July 16, 2004 Would a "deadman" failsafe system allow an automatic system to alert medics/paramedics and seek cover/return to bay?   Speaking about mil/construction use there, sorry it don't make sense... -- Posted by Ajay on 12:37 pm on July 16, 2004    The military is already working on AI for autonomous robots.  The Department of Defense (DOD) recently held a competition where serveral teams made AI controlled robots that were to make their way, unassisted, across a section of desert.  Most didn;t make it more than a few miles before they stopped working or ran into other issues.  This technology still needs some maturing.    Automatic alerts if the pilot fails to respond and things like that should be fine though.    However, if the operater was somehow killed or critically injured so much so that he is no longer capable of any kind of effective movement, I'm not to optomistic regarding what the overall situation for the vehicle is either. -- Posted by cakefool on 1:15 pm on July 16, 2004 I was thinking more "overweight construction worker has heart attack and mech drives him to the medic" stories, though that may be followed by the "Dead is good: why our software raises your productivity if your workers die" product report - ooh, got to go, someone just brought round the free cholesterol flavoured doughnuts... -- Posted by cakefool on 1:17 pm on July 16, 2004 In other news - does anyone know how the japanese 'autonomous model helicopter search and rescue aid' project worked out? -- Posted by kaempfer on 6:17 pm on July 16, 2004 Have we already considered the possibility of one pilot and one person to work the arms and everything? -- Posted by ChrisDickerson on 6:39 pm on July 16, 2004 Yes. We thought about it. There are a few cases where it would make sense, like in underwater exploration or in extremely hazardous areas where a pilot and specialist are both required to complete the mission. But, for the most part, there will only ever be a single pilot/operator. -- Posted by Viking420 on 6:53 pm on July 16, 2004 A deadman switch to start the "four way blinkers" and sit the mech down would be a good idea at least. I'm not so sure where you would put it though... the cockpit specs so far seem to depict an operator held fairly securely in a stationary position. For a literal "deadman switch" you could use a hydrometer in the seat... but that's getting a little twisted even for me... *grin* -- Posted by OrionOne on 10:58 am on July 23, 2004 I'm really not sure what you'd use to detect loss of pilot consciousness.  That's what your deadman switch should detect.  If the pilot is still alive but unconscious, and the mech is still moving, he won't be alive for very much longer :P   I am not a doctor, but is there a brain activity or heart monitor or something that can, without fail, determine consciousness of the subject? -- Posted by Merlyn on 11:39 am on July 23, 2004 Usually a deadman switch refers to a physical switch that the operator has to hold down while using.  Riding lawnmowers often use a toggle under the seat to insure that the mower shuts off if you fall over (obviously not good enough). Perhaps a small safety button under the pinkie on the hand grip would work. -- Posted by OrionOne on 1:16 pm on July 23, 2004 ah, right.  I was thinking of something more along the lines of the lanyard on an ATV or Snowmobile or Jet Ski, except with a mech you can't fall off, so you'd need to figure it out another way. Maybe the operator pressing his head back against the headrest?  or maybe a strong tug on a helmet restraint....a pinky button would be a pain to hold for a long time. -- Posted by Viking420 on 2:30 pm on July 23, 2004 Maybe a magnetic bracelet or ring along with a magnetic switch on the controlls? Or a simple thermal or pressure sensor could do the same thing. Of course, if you die violently you might have muscle spasms that could just make you squeeze the controlls even harder. Then again, a violent death inside a mech would probably mean the mech itself is in trouble too... Is a dead man's switch even really needed? I can definately see SOME benefits, but depending on the final design of the controll interface a dead man's switch might be redundant anyways... one small point in favor of an interface that requires some micromanagement of mech movements is that a slumped over but stationary pilot would only move the mech once and not just unwittingly propell it in a random direction until it hits something/someone or runs out of gas. -- Posted by Drachillix on 10:26 pm on July 23, 2004 Just stumbled across this place...looks like fun!  Give me a ring when you need a parts/inventory manager. As far as detecting pilot consciousness and or condition something akin to a dynamap blood pressure monitor could easily give you several clues to a pilots condition via blood pressure and pulse.  If gloves were worn, pulse oximetry sensors could be implemented as well giving an idea of respiratory function. As for control interfaces, put me firmly in the upper body waldo camp, with a concession 1 Waldo is generally locked down and the movement joystick is is there.  Most of the time either you would not need both arms to perform the same task and if you did, they could be slaved to the same waldo.  Tandem armed grab/lift functions could easily be accomplished via hats on one waldo since you now are basically operating a bipedal clamp truck. An array of buttons at the handgrips could handle a variety of functions, voice commands for various other items. For movement controls I kinda like the idea of a joystick of some kind with forward back left right and twist to turn.  Torso twist if needed could be toggled, probably rarely used while in motion for several painfully obvious reasons. Balance and fine leg controls, sorry, not gonna happen on any regular basis.  This would almost have to be 100% computer control, just give the computer a direction and let it do the walking, rough terrain would be handled by slowing down and taking smaller more stable steps to avoid falling.  Stability would be rule #1 and our little brains are not up to the job that gyroscopes and computers can do. For some control ideas, many folks should take a look at newer models of forklifts and especially clamp trucks.  They share many of the problems of mecha and are powerful, heavy, and stability under load is a constant concern.  To the end user, the mechanics of locomotion should be transparent.  We don't think about walking, we just do it.  If the pilot has to constantly focus on movement, he wont get any work/fighting done. -- Posted by cakefool on 3:41 am on July 26, 2004 My thoughts for the deadman switch (yes, physical) were more of the:  "I'm interested in buying your construction walkers, but    i'm a bit of a technophobe - will they go crazy and    trample everyone to death if the operator has a    seizure?"   "No sir, they have a deadman switch - any problems like    that and they stand inactive - they can even call for    help if the pilot is disabled..." See my point?  More of a selling point than a valid feature, and how many automotive companies have put that kind of thing in just to help sell? (I'm thinking Volvo/cupholders) Welcome to the forum Drachillix -- Posted by Drachillix on 11:22 pm on July 27, 2004 "No sir, they have a deadman switch - any problems like that and they stand inactive - they can even call for help if the pilot is disabled..." See my point?  More of a selling point than a valid feature, and how many automotive companies have put that kind of thing in just to help sell? (I'm thinking Volvo/cupholders) Actually its is an extremely valid feature, take a look at the world of forklifts.  An injured or impaired forklift driver can kill/crush people and merchandise with alarming speed.  A runaway loadermech could easily drop expensive cargo or even walk off a dock into a harbor trashing the mech as well as its pilot. In my example of pulse/blood pressure monitoring it could shut down the mech if pulse and or BP exceeded certain ranges. Welcome to the forum Drachillix Thank you  I only live a few hours from MPS I hope to visit someday if only out of curiosity. -- Posted by cakefool on 4:09 am on July 28, 2004 Sadly this restricts MECHonstructors to areas without billboards - imagine the havoc causes by the reactions of the pilots to the new victorias secrets poster... technically I only live a few hours from MPS too - I just have to get Concorde back in the air and charter a flight to the MCF - reckon the parking lot will take it? -- Posted by geneofols on 11:31 pm on July 28, 2004 Well, it wouldn't do much if they were incapacitated, but to keep the operators from getting out and leaving the thing running, you could give it someting like a lot of lawn mowers have these days; a kill-switch under the seat.  It won't run unless there's enough weight on the seat to compress a small spring and close a contact. -- Posted by Ajay on 3:46 am on July 29, 2004  Could the vehicle stop walking/ligting/etc when the operator is no longer providing input for the action? I mean I understand the potential usefulness of some kind of safegaurd, but if the operator is seriously impaired/injured- wouldn't that imply some situation alot more grave than the vehicle swying around for a few seconds before comming to a standstill?   For instance, what kinds of situations seriously cause a vehicle operator strapped into a large construction vehicle to suddenly fall unconscience or go into uncontrollable spasms?  Also, what would be the consequences of such a situation on the vehicle, surrounding vehicles, and personnel (aside from the operater losing control)?  I'm personally stretching my imagination for this one.  Most that I can see would involve damage to the vehicle and surrounding personnel in and of itself, or be the result of poor/unsafe work standards.     <>     So I take it that loss of control through operater incapacitation is common in current construction/engineering vehicles as well?  What kind of safegaurds do they use? -- Posted by cakefool on 5:24 am on July 29, 2004 I wrote a big long answer to that, but it turned out to be unreadable rubbish when I previewed it, so here it is in short -   Pilot slumping over controls with stroke/seizure/heart attack may not be common, but can happen, and when many lives may be at stake it makes sense to have some safguards in place. -- Posted by cakefool on 5:29 am on July 29, 2004 I'm wasting everyones time here, that was no answer. Most safeguards i'v seen are of the form - keep endangered personnel out of dangerous areas, protect valuable cargo with elaborate packaging.  None seem to address the issue we are discussing -  don't allow the situation in the first place - other than small scale robotic applications -  if something is in the way, don't proceed, if something does/doesn't happen when it shouldnt/should, alert controller. -- Posted by Ajay on 9:32 am on July 29, 2004 <>   How often and under what circumstances? <>  How often does this lead to actual damage to surrounding personnel and systems? <Most safeguards i'v seen are of the form - keep endangered personnel out of dangerous areas, protect valuable cargo with elaborate packaging.  None seem to address the issue we are discussing - don't allow the situation in the first place>>   I've never heard of such a thing being much of a problem.    Not hiring people who are prone to heart attacks and siezures, avoiding things like substance abuse and fatigue, hardening the crew compartment to ensure the operater is not struck, and providing control systems that do not begin operating autonomously when the operater stops moving, etc.  These all seem like ways of preventing damage to surrounding personnel and systems due to operater incapacitation- without something hooked up to monitor the operator's vitals.     If what we're worried about here are freak accidents, then I don't think the addition of the system will be putting anyone at ease as much as it is making them worry. Also, when taking into account that it is yet another item that must be placed into the crew comparetment, trained to be used by the operater, undergo maintainence to make sure it works properly, and adds money to the price tag and operational costs of the vehicle- you can see why someone would wait until they felt they could really benefit before applying the system. -- Posted by Drachillix on 9:48 pm on July 29, 2004 if the operator is seriously impaired/injured- wouldn't that imply some situation alot more grave than the vehicle swying around for a few seconds before comming to a standstill? An unguided piece of heavy equipment is dangerous to many others, not just the operator. For instance, what kinds of situations seriously cause a vehicle operator strapped into a large construction vehicle to suddenly fall unconscience or go into uncontrollable spasms? Tired, overworked, heat exhaustion, dehydration.  Construction sites are famous for being lacking in some basic infrastructure like running water or shade. Also, what would be the consequences of such a situation on the vehicle, surrounding vehicles, and personnel (aside from the operater losing control)?  I'm personally stretching my imagination for this one.  Most that I can see would involve damage to the vehicle and surrounding personnel in and of itself, or be the result of poor/unsafe work standards. Construction and lift equipment is often used in very close quarters and in many cases items are being manually unloaded by people around the lift area.  In these types of conditions so much as scooting forward 6 inches could easily pin and or snap someones ankle without the operator even feeling it.  Forklifts also operate in environmnets where they only have 6-18" longer than their length to turn around in.  A slip or stumble could mean taking out storage racking or part of a structure people are working on.  knocking people off and or causing potential avalanches of materials or merchandise.  In case you are not aware of it, a typical little forklift you see zipping around weighs about five tons.  Heavier lifts easily weigh in at 15-20.  I don't really see how construction mechs of any useful size will be much lighter. So I take it that loss of control through operater incapacitation is common in current construction/engineering vehicles as well?  What kind of safegaurds do they use?  It happens, its uncommon, but it happens.  Standup forklifts all use some kind of deadman switch, sitdowns have a seat switch last I heard. Some of these features prevent the unit in question from proceededing if a rider is jarred loose by a collision or incapacitated in any way.  Generally speaking its not to protect the operator, its to protect others.  The operator is well protected, people around them are often not.  In any workplace accident, policy tends to make anything that happens the vehicle operators fault no matter how drunk or stupid the person he clobbered was. Oh, and items are usually packaged to prevent damage under normal handling, not under long fall or crushing force situations.  More padding = bulkier = more expensive to ship = more costly goods. -- Posted by Ajay on 3:49 am on July 30, 2004    Before I begin I'd like to say I like the idea of making the vehicle stop when the operator does, but I don't think one needs bio-monitoring to this- as poosed to simply letting go of the way he provides input. <>  I've acknowledged that. <>   These lead to unconscious and uncontrollable spasms?  I just thought that these situations often led to impaired perception and poor control of motor-skills.  This causes the operater to make poorer jusdgements- but most of the time he is still in total control of the vehicle.      So the bio-monitor would decide whether or not it's operater were unfit irregardless of their current capacity at the time and shut itself down?  I would be more inclined to let the operater simply stop providing input when he is no longer capable of continuing.  A small switch or something on the joysticks that could be held down could accomplish this.    Such a situation is also the result of the poor safety standards I mentioned earlier.  I suggest that more time, money, and resources be spent on preventing such circumstances, as opposed to incoorporating more complex systems to monitor the pilot in case he succumbs to the above. <<  It happens, its uncommon, but it happens.  Standup forklifts all use some kind of deadman switch, sitdowns have a seat switch last I heard.>>   What kind of dead man swithces though?  Certainly not or bio-mointoring equipment I would think.  Most seem to be directly controlled by the operater more or less.   Most that I've seen, feature a switch of some kind, that when released, causes the vehicle to stop moving.  Essentially the same as the operater simply no longer providing input into the machine.   (Edited by Ajay at 8:40 am on July 30, 2004) -- Posted by Earle Bishop on 8:45 am on July 30, 2004 <> I am pretty sure OSHA prevents that kind of thing. I cannot speak for other states, but California's highway maintenance system "CalTrans" has its workers stopping to rest about every 15 minutes. They mostly work at night, too. I work the guys hard here, and I mean hard and we have not once had anyone pass out or loose control of a machine or their load. A deadman switch is almost infallible. Bio monitoring equipment is not. -- Posted by ChrisDickerson on 2:34 pm on July 30, 2004 I'd say keep it simple. A wrist strap connected to the chair arm by a short coil-cord. The cord runs to the computer/health monitor. If the pilot's heart rate gets above or below a certain point, then the system automatically brings the mech to a halt and sends a radio call to its home base for medical assistance. -- Posted by Earle Bishop on 5:33 pm on July 30, 2004 That might work in an airconditioned, enclosed military cockpit, but it will not pass for the standard sweaty CalTrans worker that gets in and out of his vehicle, or must contort to see what he is digging at. A few other items that come toi mind are: How would you affix it comfrtably so it never looses signal? How long before that coil cord gets notted or twisted like a phone cord? What if it disconnects during limited flight? What if the machine is at a full run, and then jumps, causing the pilot to momentarily black out while leaping over a ridge... *splat* What if power drops to the life support system, it gets bad data, or the pilot has a seizure (given, most people with known seizures do not have vehicle operator's permits)? -- Posted by EdZ on 6:02 pm on July 30, 2004 If you divided the two kinds of movement of the mech into categories, active and scripted (where active needs constant imput from the pilot, and scripted is computer-controlled), then if the pilot is incapacitated in some why, then active input would cease, but passive movement (i.e. a construction worker has just instructed his mech to walk to the other end of the site) will comtinue. As long as all passive movement has a definite conclused programmed in (i.e. you can tell the mech to 'go there' or 'go 100 meters that way' but not just 'go that way') and the computer can detect obstacles in it's path and track other moving objects, I can see no situation whre the mech could injure someone or cause damage with no pilot input. (Edited by EdZ at 3:03 pm on July 30, 2004) -- Posted by Drachillix on 8:03 pm on July 30, 2004 A few other items that come toi mind are: How would you affix it comfrtably so it never looses signal? How long before that coil cord gets notted or twisted like a phone cord? Short range say (4-6") RFID device on a bracelet -- Posted by Drachillix on 8:30 pm on July 30, 2004 Quote: from Ajay on 12:49 am on July 30, 2004 These lead to unconscious and uncontrollable spasms? Heat stroke will normally result in some kind of seizure activity. I just thought that these situations often led to impaired perception and poor control of motor-skills.  This causes the operater to make poorer jusdgements- but most of the time he is still in total control of the vehicle. Total is a mighty strong word.  I have seen forklifts running hard on 100+ degree days in non-climate controlled warehouses and docks.  Plenty of people are tired and sweaty to the point of minor impairment, this is not unique to any one company or area, it happens all over. So the bio-monitor would decide whether or not it's operater were unfit irregardless of their current capacity at the time and shut itself down? Biomonitoring would cover a wide range of issues that would not be as sensitive to damage or bypassing like taping down a button or an object crushing/penetrating the cockpit and pinning the operator in place. I would be more inclined to let the operater simply stop providing input when he is no longer capable of continuing.  A small switch or something on the joysticks that could be held down could accomplish this. This depends on how the controls are oriented.  If waldo devices are in use it becomes more difficult to determine if a motion is an operator passing out or an intentional movement. Such a situation is also the result of the poor safety standards I mentioned earlier.  I suggest that more time, money, and resources be spent on preventing such circumstances, as opposed to incoorporating more complex systems to monitor the pilot in case he succumbs to the above. Whats cheaper in the long run, putting the feature on the mech, or additional training for everyone who works around them. As a kind of alternative based on my prior post an active RFID system with sensors in the legs and the ID chips say in the workments helmets could easily trigger some kind of warning that people are working near your feet. What kind of dead man swithces though?  Certainly not or bio-mointoring equipment I would think.  Most seem to be directly controlled by the operater more or less. Not that I am aware of. Most that I've seen, feature a switch of some kind, that when released, causes the vehicle to stop moving.  Essentially the same as the operater simply no longer providing input into the machine. Many of which could continue to provide input from an operator slumped over the controls. We seem to be spinning this in much different ways.  I compare worker mechs to forklifts as a minimum example.  You seem to be operating under the assumption that if a forklift does not need it, then a mech does not. Mechs will be far more complex and in many ways far more dangerous due to the additional flexibility of their arms and legs.  Forklifts dont sidestep and have much better visibility than any mech cockpit arrangement I can think of.  If anything workermech safety should be far more stringent than forklifts. -- Posted by Drachillix on 8:36 pm on July 30, 2004 Quote: from Earle Bishop on 5:45 am on July 30, 2004 I am pretty sure OSHA prevents that kind of thing. I cannot speak for other states, but California's highway maintenance system "CalTrans" has its workers stopping to rest about every 15 minutes. They mostly work at night, too. I work the guys hard here, and I mean hard and we have not once had anyone pass out or loose control of a machine or their load. I have seen 2 accidents, both on the tail end of a hot summer day.  Both times operators were hot and sweaty to the point of difficulty seeing.  Should they have stopped earlier, yes, were they showing good judgement in their state, no.  If you can't make the decision to stop, you can't make good decisions about doing your work. Both of the people in question were fired, but it did not stop the accidents from happening. -- Posted by Ajay on 5:18 am on July 31, 2004 <>   By "total" control I was implying that the operater is still supplying the vehicle with input.  The vehicle is not going crazy because it is operating on it's own, but because the operater is supplying the necessary input. <>   However, neither is a physical switch as subject to as wide a range of misreadings, is overall much simpler to implement, and seems to already be performing the job adequately. <>   This is certainly an issue that should make us wonder to what degree and in what situations waldos should be used. <>   I am not questioning dead-man switches in themselves.  I'm challenging bio-monitoring ones, and suggesting a simpler physical switch as a better alternative.    Monitoring your heartbeat and bloodpressure is easy.  Checking an increase in heartbeat is easy.  Telling the difference between an increased heartbeat.blood pressure, sunstroke, a heart attack, nervousness, stress, etc, from just the heartbeat and using that to decide whether or not the operater is capable of continuing is not easy.   Monitoring whether the operater's perceptions are altered due to fatigue is not easy using only the heartbeat and bloodpressure.  Monitoring whether or not the operater is conscious via a heart beat or blood pressure is not easy. We could add in systems to mointor body temperature, but this would require placing the sensor in some uncomfortable place (mouth, armpit, etc) in order to get accurate readings.  We could add in some kind of system to mointor the nervous system, but this would require multiple clean contact points with the body.    Throw in the fact that many people have different vital signs and react to different situations in slightly different ways, and we have a system that's not so much more effective than a physical switch, but much more costly to implement.  However it is not only much more expensive, but much bulkier, and requires much more training and acomodations by the operater.   In either case, increasing worker safety need not go as far as equipping every worker with a high tech gadget.  I would think that any time a group gets a totally new class of machine, they at least familiarize those who are working right beneath it about it's basic capbilities and any safety hazards it could create, and any possible ways of avoiding/mitigating such hazards. <>   As long as their hand/finger remains on the switch(es) of the physical deadman device.  Lack of consciousness often leads to a loosening of grip though. <>    That's the general idea.  I'm not against innovation, but when looking at similar systems with similar functions and potentially similar systems, I would think that there would be other traits/systems that could be traded between the 2.   If we are going to suggest the use of new experimental systems, I'd really like to see them tested in pre-established vehicles undergoing the same general rigors, before suggesting they become standard part of something that doesn't even exist yet. <>   Flexibility in arms and legs can be relatively simple depending on design, but I agree with the general statement.  However, I don't think bio-mointoring equipment is really much progress towards a safer work enviroment compared to alternatives. <Both of the people in question were fired, but it did not stop the accidents from happening.>>  Would/could a bio-monitor (or any kind of deadman switch) have prevented these particular incidents?   (Edited by Ajay at 6:39 pm on July 31, 2004) -- Posted by noahrei on 10:03 am on July 31, 2004 Though you don't here of those sort of driver related accidents very often they do happen. I have seen the carnage when the drive of a full size delivery van had an epileptic seizure at the wheel while he was approaching a redlight at about 45 mile an hour. The car he ran over looked like a sardine can...and so did the two ladies who were in it... As much as I'd like to see some sort of system to prevent that, I find it hard to think of ways that the vehicle could know what is happening. -- Posted by Drachillix on 10:37 pm on July 31, 2004 Throw in the fact that many people have different vital signs and react to different situations in slightly different ways, and we have a system that's not so much more effective than a physical switch, but much more costly to implement.  However it is not only much more expensive, but much bulkier, and requires much more training and acomodations by the operater. You would be looking for variations in vital signs rather than absolutes.  The monitoring system can set a rolling  baseline for the pilot, it could trigger an alert or reduction in  and only shut down if it varies say more than 20% from baseline. Temperature could actually be easily derived at the same sensor as Blood pressure.  Its just a matter of establishing normal for that spot on a persons arm and varying from there. In either case, increasing worker safety need not go as far as equipping every worker with a high tech gadget. Maybe you don't understand RFID, this "high tech gadget" is the same thing that door alarms use at stores.  RFID detection arrays would be pocket change in the cost of a mech, the chips themselves for workers are about $0.25 each. In retrospect I could see where this RFID idea might work better than biomonitoring overall.  A person within say 15 feet of the legs would slow down the mechs motion to be less likely to smash a co worker.  Further slowdowns could be triggered if more RFID tags enter the monitored area.  You might even be able to make it possible for a group of say 10 or more people to stop a mech if they can get enough sensors near the feet. This way, there is zero training involved.  You wear your helmet anyway on most construction sites where mechs would be useful. You could also easily make enforced slowdown areas and or a "mechstopper" that is just a cluster of tags bound together or stuck on a wooden block that you could toss at a misbehaving or erratic mech pilot to stop him. With a little work this could be applied to forklifts and it might be a good idea there (forcing speed reductions for blind corners or in the presence of other nearby lifts. Would/could a bio-monitor (or any kind of deadman switch) have prevented these particular incidents? Biomonitoring, a small chance, deadman switches did not.  RFID would have prevented or reduced a minor injury suffered by somone bumped in the secoind accident. -- Posted by Ajay on 11:03 pm on July 31, 2004 < Temperature could actually be easily derived at the same sensor as Blood pressure.  Its just a matter of establishing normal for that spot on a persons arm and varying from there.>>   However there still exists no standard for humans.  While increases from the baseline can occur, it's difficult to tell from human to human whether this has resulted in an operater who is no longer able to do his work- or perhaps misdiagnosing a perfectly capable operater as being incapacitated.   In the case of internal temperature sensors, skin temperature often fluctuates by several degrees in either direction (sweat, airconditioning, wind due to movement, etc). This is why most all thermometers in medical situations are applied internally.  A change of internal temperature by 1-3 degrees can be serious.  A system measuring skin temp is not going to be a particularly effective way of mointoring actual internal temperature- which is what really counts in diagnosing conditions.   A reliable thermometer is going to need to be placed into the mouth, or under the arm.     Neither heart monitoring, nor temperature readings can measure drowsiness or fatigue as far as I'm aware of.   <>  My misconception. <>   How would bio-monitering have helped?  You mentioned that in both cases the operaters were hot and sweaty and that the incidents took place due to distorted perception.  What variable factor would the bio-system have picked up on?    I agree that a physical switch would not have helped either.  Well, at least it could let the person who's actually undergoing the phenomenah make the judgement call, instead of the machine overiding them based on much more limited input.   In either case, this would be where increased safety practices and reinforced application of safety procedures could have helped. Had either of the workers taken a short period of rest or avoided working while fatigued, the situation could have been better avoided.  By asking operaters to work under these conditions, you are most certainly asking for disaster.   To feel that you can ask vehicle operaters to perform work under lax safety conditions since their automated safety systems will prevent accidents from occuring is the beginning of a large problem- if not a large problem in itself. -- Posted by Drachillix on 12:12 am on Aug. 1, 2004 Quote: from Ajay on 8:03 pm on July 31, 2004 However there still exists no standard for humans. 98.6 degrees farenheit 72 beats per minute 120/80 blood pressure While increases from the baseline can occur, it's difficult to tell from human to human whether this has resulted in an operater who is no longer able to do his work- or perhaps misdiagnosing a perfectly capable operater as being incapacitated. Profiling operators would be simple and require very little computer work. In the case of internal temperature sensors, skin temperature often fluctuates by several degrees in either direction (sweat, airconditioning, wind due to movement, etc). This is why most all thermometers in medical situations are applied internally.  A change of internal temperature by 1-3 degrees can be serious.  A system measuring skin temp is not going to be a particularly effective way of mointoring actual internal temperature- which is what really counts in diagnosing conditions. I worked as an EMT for a few years and am currently married to a labor and delivery nurse, I am very familiar with the mechanics of vital signs and the reasons behind their application. A reliable thermometer is going to need to be placed into the mouth, or under the arm. As long as its positioning is consistent, it can be pretty much anywhere.  I pointed out the BP cuff as a likely site because it could be against the skin and shrouded from drafts or outside contacts that would give wild readings.   Neither heart monitoring, nor temperature readings can measure drowsiness or fatigue as far as I'm aware of. Nope but these conditions will both cause cardiovascular responses that could easily read out of baseline for that operator. As I mentioned before, I do agree that biomonitoring would be less valuable than I initally surmised at least on present technology.  It might be worth revisiting later. So whaddya think about the RFID triggered retarders idea?  Had either of the workers taken a short period of rest or avoided working while fatigued, the situation could have been better avoided. I misunderstood some of your prior posts, I was under the impression you wanted to train workers to be better at staying out of the way rather than better training operators to recognize their own limits.... In this I heartily agree By asking operaters to work under these conditions, you are most certainly asking for disaster. Also agree. -- Posted by Ajay on 1:23 am on Aug. 1, 2004 98.6 degrees farenheit 72 beats per minute 120/80 blood pressure  My current Temperature is about 98.5.  My heart beats at 80 beats per minute.  I have blood pressure of 130/85.  What you've listed are the healthy ideals of the avaerage majority of people- while sitting down and doing nothing.  Few people actually meet all of these to an exact match, and under working conditions, these values can fluctuate.  You've a;ready pointed out that construction work already pushes many to their potential limits, increasing the chances of someone not fitting any particular standard at any one time. <>  Then you understand that bio-monitoring sensors, while good at providing information for a person to use when making decisions, are not so good for dictating actions in and of themselves.  Especially ones affixed to more active indivuals, and when attached to less than ideal places in order to not restrict their mobility.   The vehicle should not be able to override the operater's controls because it feels the operater is incapable of fulfilling his job based on limited information that could very well be inaccurate.  The operater should decide when they are no longer capable of working- not the machine.      It seems likely that if the operater were using some kind of switch based controls, that when they begin to go into his spasms, clucthe their chest, or slump over uncoscious, they will more than likely release their grip of the hand controls, releasing the switch, and letting the vehicle's dead-man system go into play.    No muss, no fuss.  No bulky or expensive heart monitoring equipment.  No pre-registering vital signs or concerns with fluctuations.   <>   I would appreciate it alerting the operater and letting them make the adjustments rather than letting the machine decide and generally sprining it on the operater, but otherwise it sounds fine. -- Posted by ChrisDickerson on 5:16 pm on Aug. 1, 2004 Good ideas, guys. Logical arguments, as well. I like the direction this is going. I like the bio-monitor idea and the dead-man switch. What I had started thinking was that it be incorporated into a touchpad on the joystick. The deadman switch will actually cut the mech's walking/running movement (gradually, of course). The integrated bio-sensor will provide a detail of heart-rate and temperature before the deadman switch is released. If the individual pilots all have data card with them for the movement of the mech, which is planned, I see no reason to not include a baseline reading on that same pilot. Since all the maneuver data will be developed through training in a simulator, the medical data can be gathered at the same time. That way the system know when the operator is stressed to an extent that is greater than normal, or when his vitals are too depressed for the situation. I think a flucuation of 20% is a good guide-line. We need to have a medical professional on staff if something like this is going to be explored. -- Posted by Earle Bishop on 5:59 pm on Aug. 1, 2004 98.6 degrees farenheit 72 beats per minute 120/80 blood pressure Well then, I run hot (genetic), my heart beats slower (healthy), and my blood pressure is WAY higher. My heart rate skyrockets when I work out. It does the same when I get fight or flight. First time I get an adrenalin burst, my mech shuts down? I hope that the load is balanced. Heres a simple thought: This project roots and survives from existing, proven technology. Why not apply the same rules that exist for cars and other vehicular transit? If I am stupid enough to operate a vehicle unsafely, it is my own fault. I mean, the first time the damn thing shuts off on me, I am going to tie a little string around the deadman switch. I will probably not be the first or last to do that. -- Posted by Earle Bishop on 6:07 pm on Aug. 1, 2004 98.6 degrees farenheit 72 beats per minute 120/80 blood pressure Well then, I run hot (genetic), my heart beats slower (healthy), and my blood pressure is WAY higher (mech building lunatic). My heart rate skyrockets when I work out. It does the same when I get fight or flight. First time I get an adrenalin burst, my mech shuts down? I hope that the load I am carrying is balanced. I hope the misplaced construction worker down below me that triggered the heart rate increase sees what is going on and gets the hell out of the way. Heres a simple thought: This project roots and survives from existing, proven technology. Why not apply the same rules that exist for cars and other vehicular transit? If I am stupid enough to operate a vehicle unsafely, it is my own fault. I mean, the first time the damn thing shuts off on me, I am going to tie a little string around the deadman switch. I will probably not be the first or last to do that. You see, me and safety equipment do not get along. A glove will cause me to loose my grip. A set of goggles will fog up on me. A blade guard will cause me to have to force the saw along instead of letting it glide. Not once has a lack of such 'safety eqip' presented me with injury. Maneuver data is simple to grab. Defining a 'situation' for the computer to understand and mark vs. the human's vitals is very, very hard. I think Ajay worded it very well: <> In no part of this system, is the decision of the computer treated as absolute fact. Computers make mistakes. Very very accurate mistakes, but they are only as good as the information and capabilities you present them with. Maybe we should just put a watercooler in the cockpit? <> This is the most prudent course of action, preferably someone in the capacity to make certified OSHA recommendations. As is mostly the MPS credo, we research well and with good resources. I would trust a certified, proven, experienced opinion over our collective conjecture. -- Posted by Drachillix on 9:56 pm on Aug. 1, 2004 Not once has a lack of such 'safety eqip' presented me with injury. Yeah, but you have a clue. I am one of those people, probably like you, who think rules are to keep the stupid people from doing something stupid.  Unfortunately its impssible to idiot proof anything, I know people who could probably stab themselves to death with a beach ball. Maneuver data is simple to grab. Defining a 'situation' for the computer to understand and mark vs. the human's vitals is very, very hard. I think Ajay worded it very well: <> This is incorrect, maybe you have heard of automatic defibrillators? I know that something like this would probably be one of the last things that you are worried about in the grand scheme of making a working mech.  Even I agree that engineering wise you have plenty of major fish to fry before getting to it.  Even I think that there might be better ways.  However, don't underestimate the world of medical monitoring.  There is some very smart monitoring systems out there We research well and with good resources. I would trust a certified, proven, experienced opinion over our collective conjecture. Checks EMT cert....work history..... What level of licensure did you want?  I am very familiar with the relevant physiology. -- Posted by Earle Bishop on 10:46 pm on Aug. 1, 2004 Quote: from Drachillix on 6:56 pm on Aug. 1, 2004 Yeah, but you have a clue. Unfortunately its impssible to idiot proof anything, I know people who could probably stab themselves to death with a beach ball. Natural selection? Honestly, I see where you are coming from. I just hate to see a machine making the decision for you. I never looked at it like I had a clue, but I can again see where you are coming from. Most of the guys here swear by the safety equip. Just not my cup of tea. This is incorrect, maybe you have heard of automatic defibrillators? I am going to presume that these are used in hospitals, where you go when you are worried about dying. Detecting zero pulse is no big deal. Accidentally shocking someone without cause would be a very big deal, methinks. I am curious, are there any such systems deployed in a work area? Examples would be very neet. Oh yeah, I forgot to mention, one of my concerns was; I was born with Wolf-Parkinsen-White. WPW makes your heart rate dramatically increase, sometimes without cause. I was a bit young to drive forklifts, but it never affected me in any sort of day-to-day operations until I had to go get it fixed. I would feel horrid if I made a machine that was trying to be helpful, and killed someone. I know that something like this would probably be one of the last things that you are worried about in the grand scheme of making a working mech. Well, yes and no. Not a big part of the overall system, but it IS important that we evaluate stuff like this to make sure that we are not developing a system blindly. It would be a lot of work to have to go back and redesign a system to be capable of being outfitted with the stuff we are discussing here. However, don't underestimate the world of medical monitoring.  There is some very smart monitoring systems out there I underestimate nothing with monitoring capabilities. I underestimate the ability of a human being to write a program to interpret that data into real world applications. Even the mech has a kind of 'uberintelligenz,' a master process that reads results, and prevents the system from acting upon it if it looks stupid. Stupid being, 'out of range'. You get a LOT of crappy data from sensors. Checks EMT cert....work history..... What level of licensure did you want?  I am very familiar with the relevant physiology. We really cannot take any more interns on at this point. I would be interested, however, in getting a prepared document from you outlining commercially available systems that have a good history behind them. -- Posted by noahrei on 11:01 am on Aug. 2, 2004 I know people who could probably stab themselves to death with a beach ball. Fortunately those people will never touch a mech cockpit. -- Posted by Ajay on 11:13 am on Aug. 2, 2004 <>  Current designs only apply the shock when the heart is beating very fast (approaching 130+ beats per second) or incredibly low.  The closer you get to the baseline heartneat, the more likely you are to accidnetally administer a shock due to something like resting or more conventional athletic activity or relatively minor situation. ........................................................    The chances of going totally unconscious are already slim.  Some members on this board keep asserting that it can  and has happened, and that a bio-medical monitoring system would be a good way of avoiding this potential issue.  However, all testimonies so far have involved situations where the operater showed poor judgement.  Other testimonies have even suggested that the situation is even more far-fetched than initially suggested due to modern standards in safety regarding such work.   The odds of the operater comming under conditions that would trigger lack of conssiousness or life while operating his vehicle under what could be considered normal conditions is very very low.    Even then, the manual system would be just as practical in the same situation, unless someone who was undergoing a heart attack or stroke, or comming under severe heat fatigue felt it'd be a good idea to try and continue working their machine instead of letting go of the controls until they finally succumb.   Not only does this create the dead man effect when the operater is incapacitated, it also allows the operater to activate the same effect at his liesure, whether he wants to take a short breather, whipe his face, readjust himself in his seat, or perform some other action- even though his vitals are seemingly okay.   -- Posted by Merlyn on 11:33 am on Aug. 2, 2004 I for one really like the RFID idea.  If all it takes to stop a mech is to toss a small transmitter near it's feet then I can see that being very handy in admittedly rare circumstances.  And while I agree that having the system override the pilot is a bad idea, I think that perhaps having it reduce the max. speed of the mech (and it's assorted manipulators) under certain conditions would be very beneficial. Probably the best way would be to have a sliding plateau in the control-response curve rather than trying to scale the input.  That way it has no effect on the mech's operation unless it is operating faster than the cutoff velocity, at which point speed would be reduced to the new maximum and the pilot would get an idot light telling them that they were moving faster than is estimated safe for the conditions.  When the cutoff is removed it would need to allow a gradual increase in speed back up to where the pilot has the throttle positioned.  Alternately, the cutoff could physically force the throttle control to where it needs to be when it is applied, and when removed the pilot simply places the throttle back where he wants it. -- Posted by Drachillix on 12:41 pm on Aug. 2, 2004 Quote: from Earle Bishop on 7:46 pm on Aug. 1, 2004 I am going to presume that these are used in hospitals, where you go when you are worried about dying. Detecting zero pulse is no big deal. Accidentally shocking someone without cause would be a very big deal, methinks. Heh, a fellow social darwinist :D Actually Auto-defibrillators are designed for use in situations where no skilled or minimally skilled operators are present like shopping malls, amusment parks, etc.  They use (IIRC, heavy math guru I am not) a math concept called fourier transformations to do a waveform analysis and compare it to a wide range of stored cardiac rhythms. They are capable of differentiating between needs for defibrillation and cardioversion and in some cases can provide pacing for erratic heartbeats. I am curious, are there any such systems deployed in a work area? Examples would be very neet. I know Wildwater Adventures at Clovis Lakes in fresno has one. I will poke around for other workplace example.  Supposedly there are thousands of units in service.  In a medical setting they are not needed, they have people who can do it manually better than the machines in most cases. I would be interested, however, in getting a prepared document from you outlining commercially available systems that have a good history behind them. I would be honored....consider it done. -- Posted by Drachillix on 1:09 pm on Aug. 2, 2004 Quote: from Merlyn on 8:33 am on Aug. 2, 2004 Probably the best way would be to have a sliding plateau in the control-response curve rather than trying to scale the input.  That way it has no effect on the mech's operation unless it is operating faster than the cutoff velocity, at which point speed would be reduced to the new maximum and the pilot would get an idot light telling them that they were moving faster than is estimated safe for the conditions. This is kinda what I was envisioning, thank you for elaborating where I generalized. If we really wanted to reach a bit, by making the recievers directional and separate recievers in each leg, you could even "biangulate" the approximate location of the people around him.  Even if its just a "behind to the left" it could assist an operator in avoiding an unfortunate situation, even if squishing the worker in question would have cleaned up the gene pool a bit. -- Posted by Jagarus on 4:10 pm on Aug. 2, 2004 I guess I should put something here since i actually plan on being a pilot for the mech, hey thats what im inturning for. Anyway as far as safety goes, do relize I am a Marine, just make sure there is an ejection system and a first aid kit, if im dead im dead. theres nothing else to be said about it. Maybe intergrate a "bio sensor" system just to see if im flatlined for an auto shutoff protocal. Last thing i want to worry about is wondering if the mechs computer system thinks im dead and shuts of in the middle of working with it. All in all the primary focus I want and would expect any pilot is piloting, letting your mind wander is what cause pilot related accidents. But again last thing i want is a computer that has a fixed pattern monitoring a human body which never acts the same with diffrent factors, starting a protocal that shutsdown the mech. Im not trying to shoot down any ideas, they are all good, make them simple, for our goal right now is to get a running prototype out all the "improvements" can be intergrated in later through the discretion of the core team members of MPS. (Edited by Jagarus at 1:13 pm on Aug. 2, 2004) -- Posted by Drachillix on 4:36 pm on Aug. 2, 2004 Quote: from Jagarus on 1:10 pm on Aug. 2, 2004 But again last thing i want is a computer that has a fixed pattern monitoring a human body which never acts the same with diffrent factors, starting a protocal that shutsdown the mech. [bows humbly before the worlds first mech pilot to be] I had mentioned before that the patterns could be keyed to the individual pilot and are not fixed at all and that they would float with subtle variations to a point.  Significant variance from your personal profile/normal range would be the preferred trigger.  Perhaps each pilot could carry a "key" like a compact flash card that has all of his preferred control configurations, bio profile, piloting hours, maybe even password/thumbprint/voice sampling info if such security became needed.  I'm sure there will be plenty of personalized info that will be helpful in the long run. -- Posted by Viking420 on 3:40 pm on Aug. 4, 2004 I have a perfect example of a vehicle with integrated throttle/controlls and a dead mans switch. A bicycle will not keep moving forward with an incapacitated operator because its operation depends on constant rythmic input by the rider's legs. With balance for a mech controlled by a computer rather than directly by the operator there is less danger of the mech falling when the operator "slumps over". I realize the drawbacks of requiring a seperate operator movement for each step, but I believe it can be arranged in a relatively non-tiring way and in the end will provide more flexibility, as well as an integral "dead mans switch" system. If the pilot is incapacitated he will give one final instruction, probably a slump forward into the joystick controlls and a push forward on the pedals. In a throttle based system this will cause the mech to walk or run forward  unless a dead man's switch is used. In a system that requires the operator to make a new movement for each mech movement this will result in a single step forward likely followed by a very strenge stationary posture for the mech as it struggles to keep itself upright while accomodating the strange input it is recieving from the operator. If the pilot goes into convusions for whatever reason the mech will probably fall over and wiggle on the ground as the balancing routines are overpowered and confused by the erratic inputs from the operator. To put it simply, requiring the pilot to concentrate and provide detailed input for the mech to operate will prevent the mech from functioning when the pilot is no longer capable of doing so. -- Posted by Drachillix on 5:58 pm on Aug. 4, 2004 Quote: from Viking420 on 12:40 pm on Aug. 4, 2004 To put it simply, requiring the pilot to concentrate and provide detailed input for the mech to operate will prevent the mech from functioning when the pilot is no longer capable of doing so. We I thought it has already determined however that a person is not capable of dealing with the bazillions of variables for foot/ankle/knee positioning to effectively walk a mech via direct control. Crazy out of left leg contol idea. In the book "Disclosure" there was a VR sytem that you were harnessed over a platform of balls, kinda like walking across an array of tightly packed trackball mice creating a unidirectional treadmill.  If a mech started to tip we should instinctually place our feet in such a position to catch or correct our tippage.  If the mech stepped to place its foot in the same relative position, could it effectively balance or would it be too slow to respond? The cleaning nightmares would probably be ugly too. -- Posted by Viking420 on 11:46 am on Aug. 5, 2004 I am not advocating direct manual controlls for balance. The continuous adjustments of position in the ankles, hips and waist are best left to a computer. I believe I have a rough cockpit design that COULD allow manual balancing of the mech, but I don't think it would be practical for a machine as large as the Mk 1. A smaller vehicle, say a half tonne and 4 meters tall could be easily controlled by a haptic interface but that's really getting closer to an exoskeleton than a mech. A computer assisted haptic interface could IMO handle the controll and balance of a mech easily. Neither the computer nor the operator would be capable of balancing a moving mech independantly but together could easily handle the task. I suggest balancing sensors, computers and actuators built into the ankles, making the rapid fine adjustments to the foot's orientation to maintain balance. Combined with a controll yoke linked to the torso's attitude, this could allow a proficient pilot to easily manage all the variables involved in dynamic balancing. The key word is a PROFICIENT pilot though, this is a complex and highly specialized vehicle concept and frankly not everyone will be suited to operating it. For example, there is no way I'll ever fly an F-16, my eyesight isn't good enough. -- Posted by Earle Bishop on 4:11 pm on Aug. 5, 2004 Quote: from Viking420 on 12:40 pm on Aug. 4, 2004 To put it simply, requiring the pilot to concentrate and provide detailed input for the mech to operate will prevent the mech from functioning when the pilot is no longer capable of doing so. Exactly, it will no longer function. Then the machine will fall over and crush anything under it, break itself, and possibly harm the pilot. Quote: from Drachillix on 2:58 pm on Aug. 4, 2004 We I thought it has already determined however that a person is not capable of dealing with the bazillions of variables for foot/ankle/knee positioning to effectively walk a mech via direct control. It has been determined, and explained to Viking. He chooses to disregard it. Quote: from Viking420 on 8:46 am on Aug. 5, 2004 this is a complex and highly specialized vehicle concept and frankly not everyone will be suited to operating it. For example, there is no way I'll ever fly an F-16, my eyesight isn't good enough. Speak for yourself. MPS intends to make a machine that can be operated by almost anyone, of almost any body size. Just like a bike. MPS has Neural Networks planned for use in several ranges of operation, and monitoring several items in tandem, including; ground pressure expectations, automated balance, automated balance while carrying a load. Heck, several of our systems are passive. You cannot break passive mechanics. It does not malfunction. -- Posted by Ajay on 4:35 pm on Aug. 5, 2004   In reference to a dead man switch, may I ask what are the potential shortcommings of a trigger sitting on the controls that must be constantly held down (like a safety/arm) to allow the operater to continue to provide input? -- Posted by Merlyn on 4:53 pm on Aug. 5, 2004 Probably comfort, as the pilot must be capable of keeping the switch activated the entire time he pilots the mech. Also, what if he needs to move his arms to reach a control or switch?  Or take a drink from his canteen?  He can't keep his hands on the controls the whole time. Think how uncomfortable you would be if you had to keep your hands on the steering wheel of your car, in the correct positions, or the whole thing would just die on you? -- Posted by Viking420 on 4:54 pm on Aug. 5, 2004 <> I will re-post the relevant sections of my earlier post... "With balance for a mech controlled by a computer rather than directly by the operator there is less danger of the mech falling when the operator "slumps over"." I have been converted... you guys won that argument a long time ago and every idea I have presented is my attempt to build upon the idea, not replace it. My concept is compatible (I believe) with the Neural net controll scheme, it merely gives more controll to the pilot over major movements while placing the computer solely in the role for wich it has been built and programmed. -- Posted by EdZ on 5:03 pm on Aug. 5, 2004 Quote: from Merlyn on 1:53 pm on Aug. 5, 2004 Probably comfort, as the pilot must be capable of keeping the switch activated the entire time he pilots the mech. Also, what if he needs to move his arms to reach a control or switch?  Or take a drink from his canteen?  He can't keep his hands on the controls the whole time. Think how uncomfortable you would be if you had to keep your hands on the steering wheel of your car, in the correct positions, or the whole thing would just die on you? All you have to do is stop to take a break. Seeing as you would usually wan't to (or more likely, be required to by law) stop all the work you're doing when it's time for a break, this is a non-issue. As for switches etc, you could have a switch for each hand, so you can alternate hands or just stop for a bit if you have to do something more complex (which would be best to do anyway). -- Posted by Ajay on 5:36 pm on Aug. 5, 2004  I'd like to clarify.  Taking hands off the switch wouldn't cause a vehicle to shutdown totally.  It would make you unable to deliver commands via the joysticks and foot operated controls. -- Posted by cakefool on 3:26 am on Aug. 6, 2004 Watching a docu on the channel tunnel years ago, they explained there were a pair of pedals that had to be held down for the entire journey, or the computer thought the pilot was dead.  Problems came when a pilot sneezed in the simulator, and took his foot off the pedal for a moment accidently.  The train came to a stop and refused to restart - the pilot was dead, so the pressure on the pedals could just be from his dead body.  I understand theres now a five second window between triggering and brakes for the driver to key his 'alive code' into the console.   I consider this a problem, don't know about you... -- Posted by EdZ on 4:58 am on Aug. 6, 2004 Then don't use pedals, usa a small switch in the joystick which is very unlikely to be pressed by a slumped body. -- Posted by Merlyn on 9:57 am on Aug. 6, 2004 Quote: from cakefool on 12:26 am on Aug. 6, 2004 Watching a docu on the channel tunnel years ago, they explained there were a pair of pedals that had to be held down for the entire journey, or the computer thought the pilot was dead.  Problems came when a pilot sneezed in the simulator, and took his foot off the pedal for a moment accidently.  The train came to a stop and refused to restart - the pilot was dead, so the pressure on the pedals could just be from his dead body.  I understand theres now a five second window between triggering and brakes for the driver to key his 'alive code' into the console.   I consider this a problem, don't know about you... Great example of best intentions not doing the job. So Deadman switches have their problems.  I like Ajay's version where it simply disables the main control interface. -- Posted by Drachillix on 1:15 pm on Aug. 6, 2004 Quote: from Merlyn on 1:53 pm on Aug. 5, 2004 (Edited by Drachillix at 10:17 am on Aug. 6, 2004) -- Posted by Drachillix on 1:24 pm on Aug. 6, 2004 Quote: from Viking420 on 1:54 pm on Aug. 5, 2004[br My concept is compatible (I believe) with the Neural net controll scheme, it merely gives more controll to the pilot over major movements while placing the computer solely in the role for wich it has been built and programmed. Problem being...major movements are the major balance issue.  The person controls motion in terms of direction and speed.  The mech will determine the best leg placement/strides to achieve that goal while holding the mech in balance at all times.  Having the operator be able to say, take a 3 foot sidestep to the left leg while the right leg is off the ground creates quite a "stability challenge". -- Posted by Viking420 on 4:36 pm on Aug. 6, 2004 OK, right leg is off the ground, you "lift" your left leg and the body will lower. The computer senses balance is being compromised and repositions both the right leg and the torso to maintain balance while providing appropriate feedback to the controlls to inform the pilot of the changes in body position. The computer and pilot are constantly communicating with each other through the controlls, the pilot is telling the computer what he wants the mech to do and the mech is telling the pilot how it will happen. The computer still determines what each limb and actuator does... and it will communicate this to the pilot. If the pilot disregards the computer's advice and pushes through the feedback then the machine WILL very likely fall over... but that is obviously the operator's fault snd not the machine's. And remember, this is all just an idea... and not a very refined one at that, simply a rough draft proposal for a POTENTIAL interface with lots of flexibility at the expense of SOME user friendliness. -- Posted by EdZ on 5:24 pm on Aug. 6, 2004 It seems pretty pointless to use a haptic interface when the controls will actually be non-haptic in nature. If the computer is doing the work anyway, why make the pilot do more than he has to for no extra benefit? -- Posted by Viking420 on 6:34 pm on Aug. 6, 2004 Well thats just it, sometimes the computer doesn't have the right idea. It is there to assist the operator, not do the work for him completely. It would allow much more flexibility in directing the computer. You would have more direct controll over the movements of the machine, THATS the selling point, not ease of use or safety but simply the added choice of what structure of the mech to use for a given motion. To walk or run forward you could either push the torso yoke forward, forcing the mech to put a leg forward (and feeling it happen through the controlls) or you could put the foot forward and the torso would lean forward to even the weight distribution. Its like the argument between standard and automatic transmission. Some people wonder why anyone would bother with a finicky clutch and geat shifter when an automatic can do the job just fine... some people wonder how people can give up the extra controll over the vehicle and reliability in the name of user friendliness. Like PC vs Mac, windows vs Linux... etc. My cockpit design could work as either a haptic or regular interface with the flick of a switch, its still got two foot pedals, a central joystick and probably dual thumbsticks and some buttons/rollers for arm controlls. For those who just want to get they're work done and not think about it, switch to automatic controlls and cut out the feedback. For those who want to be able to feel what the machine is doing and influence it's decisions then "manual" controlls could be activated. It really would be just a matter of preference, either way the machine is still going to do its very best to stay balanced. I'd rather go for the manual controlls myself... but then again I detest automatic transmissions :) I'm not saying either way is better, I'm just tossing out alternative ideas for the pilot to interface with the computer, the computer to machine interface remains unchanged either way. -- Posted by EdZ on 6:49 pm on Aug. 6, 2004 The problem with haptic controls is a- you neeed a way to make feedback work for an entire body, which currently ther eis not way to do. b- for it to be really useful, you need a mech with human or near-human proportions, which it is unlikely mechs will be, as they will almost certainly use the best design for their particular job, and that design will rarely/never be near to human proportions. -- Posted by Viking420 on 7:17 pm on Aug. 6, 2004 Its not a full body interface, it is three controlls with 3 axis of freedom each. One for each leg that measures x and y position and force, forward and to the side. a pedal at the end of this knee-heel "inverted joystick" measures extension of the foot, heel-toe. With that setup you have controll over a direction and a distance. For the hands you have a "flight yoke" that works like a joystick that can twist. That yoke controlls the waist of the mech. It would actualy look kinda like a "stick man" that you grip by the shoulders, attach your feet to his and start "leading around". The "stick man" does not want to fall over and will react to movements that would cause it to topple while otherwise cheerfully being led around. -- Posted by EdZ on 7:34 pm on Aug. 6, 2004 But you could do the same with pedals, and not be restricted by human joints. -- Posted by Viking420 on 8:23 pm on Aug. 6, 2004 So just x and y controlls for each foot? For simple walking around that would be fine, but if you need to crouch or raise yourself you still need another controll. The hands will likely be busy managing the arms, torso, radio, a/c, coffee cup or whatever so why not integrate those controlls into the foot controlls wich are already handling basic navigation and movement? You could actualy very easily remove the torso controll yoke and trust the nerual net in charge of the ankle, hip and waist compensation movements to keep yourself up. That would free the hands up for a standard dual joystick approach, or for any number of specialized interfaces. My original point is simply that requiring the pilot to somehow engage each footstep seperately will act as a dead man's switch in and of itself. Whether the pilot swings his feet or taps his toes to do so is irrelevant to the idea. The safety device to keep the mech from crashing into things while the operator is incapacitated is simply that when the pilot stops moving, no matter where his limbs end up the mech will not take more than two steps in any direction and will then just stand there looking awkward. -- Posted by Ajay on 3:57 am on Aug. 7, 2004 <>   Sometimes the operator doesn't either.  In either situation, this is when you usually fall over. <>   I can understand the argument, but I think the popularity of automatics vs manual transmissions in vehicles and number of windows systems in use compared to Linux sys alot- although I'm sure the popularity of Windows is more about business and less about the actual benefits of the system. The increasing trend for system autonomy at the expense of detailed control for the sake of user friendliness seems to most to be the most practical route.   The alternatives in these situations also have many differences and advantages that would not really cross over into the manual/fly-by-wire walking machine issue. <>  I would think that crouching, raising, etc would either be found on the hand controls or be activated by some other system.  Unless the vehicle is doing any more than bending at the "knees" while it's assumming a lowered stance, this shouldn't be too much harder than assumming a lower stance.   You could very well do it by adding a 3rd dimension of control to the legs.  I don't think this would justify the integration of a moe manual control method though, and given the relative infrequency in which crouching/walking movement would usually be used, I wouldn't see it having enough utility to involve engineering and building a 3rd dimension of movement into the foot controls.    I can see crouching used often when... Moving in an area with many overhead obstacles, where you're constantly switching from full height to a reduced height position instead of just staying in a reduced height position most of the time. Reaching down to pick something up (longer arms could allieviate this)... Setting down for transport and getting back up... <>   Does nothing to control the arms and upper torso.   ............................................................   The issue I'm seeing with the stick man being led around is the operator's limited field of vision.  This is method of control is used in powered exoskeletons, but the operater has full view of the ground.  The operater also has full force feedback from his own walking actions since he is still firmly planted on the ground.    When sitting 10+ feet off of the ground and looking at a  series of screens, properly positioning feet around or over obstacles, would be much more difficult.     The Mechaps balancing not only reacts to input via forces like gravity and such, but also has forward scanning systems that can detect changes in elevation, etc, and makes modifications to it's stride length, positioning of feet, etc, before the foot even hits the ground.    it's simply too much for someone so dettached from the outside situation to handle effectively.  If the the vehuicle has shown it's able to do this adequately, but the operater is unsure whether or not to trust the computer to do it, then the operater should not be in the vehicle.   (Edited by Ajay at 8:20 am on Aug. 7, 2004) -- Posted by Viking420 on 12:43 pm on Aug. 7, 2004 << The increasing trend for system autonomy at the expense of detailed control for the sake of user friendliness seems to most to be the most practical route.>> In that case, most people will choose a forklift, bobcat, etc instead of a mech anyways... I tend to think that people looking for a highly flexible walking vehicle will also be the sort that drive standards, or motorcycles for that matter :) Of course, I'm not in marketing... so who knows really, maybe do a focus group? :) -- Posted by Ajay on 1:36 pm on Aug. 7, 2004 <>   Why not?  I can think of lots of other reasons why I'd choose a bobcat for several jobs over a legged vehicle.     By the time one feels the need to request use of a legged vehicle, you should be in territory approaching the potential limits of  tracked/wheeled vehicle capabilities.   -- Posted by Viking420 on 2:40 pm on Aug. 7, 2004 <> Exactly, and people living and working in those situations are not the average "soccer moms" that would be delighted by a "hands off, user friendly" interface. IMO (and experience) the people working and living off the grid and away from the roads are EXACTLY the type of people who appreciate the detailed and involved operation of they're equipement. Management might love the idea, but then again they think a weekend of paintball will cure every social ill at the office too... I'd suggest looking at this from the end user's point of view. Keep in mind who the end user will be, it will NOT be the ignorant suburbanites with a dell PC running windows XP (or an iMac, or Presario, etc) and a Ford Explorer with an automatic transmission. Granny and Soccer mom aren't going to use mechs, so why build the interface with them in mind? Enabling as many people as possible to operate the machine is a fine goal... but frankly its the people's job to learn, its not your job to make it idiot proof... lest you attract thousands of idiots. The customer is the most irritating and frustrating kind of human because no matter how stupid and impractical they are... they're always right. -- Posted by EdZ on 3:43 pm on Aug. 7, 2004 The problem is, too much micromanagement is a very bad thing. Think of how hard it would be to operate a bucket ecavator if you had to set each cylinder pressure manually and trigger them in the correct sequence. You may get more control, but I don't know anyone who would use that system rather than letting electronics do the gruntwork and use the controls to direct that work. -- Posted by Viking420 on 4:13 pm on Aug. 7, 2004 I agree completely, but you can't just tell the excavator to "dig that way" and trust a sensor and computer to know what you mean. You need to strike a balance between the two extremes. It is simply my opinion that there is not a sufficnent level of detailed controll with the current "pedals and joysticks" method for anything but the most rudimentary tasks. You COULD just add more controlls for other specific functions but that would take up space in an already very cramped environment. I believe that making the interface as dense and complex as possible (while still retaining both safety and useability considerations) will mean that there will be fewer "upgrades" required down the road. -- Posted by EdZ on 4:27 pm on Aug. 7, 2004 Pedals and joysticks can provide almost any range of movement that the mech is capable of. A mech isn't able to exhibit the same range of movement as a human body, so using one to contro lthe other will never give effective results. And situations where a mech needs to move with absolute precision would be better handled with complete autonomous control, allowing the mech computer to place the feet in the precise positions needed to sustain balance. Even with force feedback, a human operator cannot place their feet with the same precision and foreknowledge as the mech itself. -- Posted by Viking420 on 5:47 pm on Aug. 7, 2004 << And situations where a mech needs to move with absolute precision would be better handled with complete autonomous control, allowing the mech computer to place the feet in the precise positions needed to sustain balance.>> Foot placement is only ONE of many aspects of balance, and in fact sometimes it is the least important. The orientation of the torso, the ankle and hip's reactionary movements to changing stimuli and a slew of other factors also determine balance. And in the first place, I've been saying that the pilot should ENGAGE each footstep, not controll each motion of each footstep. The entire point is to have a controll scheme that cannot be operated by a continuous motion or position of the pilot, nothing more. Like I said I don't care if the pilot swings his legs, taps his toes, wiggles a pair of joysticks or picks his freaking nose to accomplish this. -- Posted by Ajay on 10:02 pm on Aug. 7, 2004 <>  You don't have to be a "soccor mom", or poorly/less trained to benefit from a simplified  or "hands off" friendly user interface.    A hatswtich control system should be able to allow control of which foot to lead off with to a good degree.  Moving one foot control forward should initiate a a turn, which will result in one leg moving forward, and if immediately followed by the other foot control moving up, should allow for single steps. <>  What particular tasks did you have in mind that could only be accomplished through such a higher degree of micromanagement through the legs?   <>  Digging is different from walking.  Different demands, and different systems exist for the ability to allow computers to do this.  I'm pretty sure you COULD make a robotic excavator though.  The Plustech forestry walker uses a fly-by-wire interface.  The operater can manually adjust general height of the chassis off the ground and the height of the steps, but even this can be surrendered to the computer if he so desires.  Operation of the tool arm is still generally manual though. (Edited by Ajay at 9:07 pm on Aug. 7, 2004) -- Posted by Viking420 on 10:43 pm on Aug. 7, 2004 I would actualy argue that "hands off" controll schemes severely limit the options available to experienced operators. While I agree that a hatswitch based leg controll system would be more than adequate for basic manuvering and navigation, any other uses for the lower extremities like kicking, crouching or jumping would require an additional controll. This controll can either be tacked on as an afterthough or integrated into the basic controll scheme. Even if these motions are completely scripted you still need a means to activate them and I can already see a mech cockpit getting too crowded for comfortable operation. On top of that, the operators "in the field" will find all sorts of extra things for a mech to do, and different ways for them to do regular duties. By all means a "training wheel" mode with the majority of the "extra" controlls disabled would be extremely useful, but once the operators begin to learn about how the machine works they will begin to want more detailed controll over they're operation. I'm just saying don't limit yourself right from the start by dismissing the idea that a skilled mech pilot can't possibly keep up to the demands placed on him. Let the pilot make the decision to either stick with the simplified "hands off" interface or "graduate" to a more involved one for improvised actions. After all, a mech fundamentally trades efficiency for flexibility. Wheels go faster and farther than legs can but not by the same routes, so don't treat them the same. And sorry for getting pissy, I'm just sick of saying "computer" and "interface" and "automatically balanced" several times in a post and then people telling me that manual controlls will never work and it needs a computer to handle all the data. -- Posted by Ajay on 4:00 am on Aug. 8, 2004 <>   Crouching, jumping, and kicking- are these the actions that you believe cannot be effectively executed using a 2 dimensional foot controls and require a 3rd dimension to be used?  Crouching and elevation of the body off the ground has been covered. As for kicking and jumping- assuming the vehicle has the capability, and these maneuvers only result in negligible stress on the design...   Given space concerns, it's quesionable whether you  could move your legs enough to even give the illusion of a kick or jump- which would more than likely make such actions totally scripted.  If entirely scripted, they can be activated by most any kind of button, switch, lever, voice, or most any form of input and would more than likely not reaire a 3rd axis of movement to be built into the foot controls.    These maneuvers are also rather complex, and may very well require the vehicle to properly orient it's upper torso and arms in a specific way to maintain balance.  If the operator is in control of the upper torso and arms, they will more than likely have to surrender control during the action.   << I'm just saying don't limit yourself right from the start by dismissing the idea that a skilled mech pilot can't possibly keep up to the demands placed on him.>>  It's not that he can't handle it.  It's more of a question of why worry him with it when the advantages are so small, and the increase in workload that much larger.    Many pilots are MORE than skilled enough to fly aircraft without the aid of computers.  Often times, the computer cushions their actual capabilitiues, keeping them from going to their full potenetial.  Has this doen anything to limit the large scale application of software controlled avionics?    If anything, the relieving many of the mundane aspects of tasks like walking and maneuvering off of the operater allows them to concentrate more on what humkans are best at- doing the job that needs to be done.  Whether that's lifting and placing a pallet of cargo into some sort of hold, erecting a large support, or locating and firing at the enenmy- the less the operator has to worry about, generally the better he is at his main job. <>   If by "Flexibility" you mean "terrain mobility" then I would agree.  If you're using the term to describes something else, I'd have to disagree. << Wheels go faster and farther than legs can but not by the same routes, so don't treat them the same.>>    I'm not treating the same.  While wheeled/tracked systems do have differences compared to legged vehicles, they also have many similarities and are vying for the same general functions- which in themselves have common demands on ALL vehicles regardless of what they use to get from point A to point B.  I also compare legged vehicles to aircraft, especially helicopters. <>  Don't worry about.  Take note though that I don't think it'd be IMPOSSIBLE for the operator.  I think they could handle a more demanding interface, but I don't see the potential advantages outweighing the increased ammount of attention it require by any good degree.  This may make him slower, make work less effecient, and potentially just be a bigger pain than a simpler system- EVEN if something were to be sacrificed.  So far though, I'm still not really convinced that you're losing that much by making walking as automated as the Mechaps team or any seemingly succesful walking robots have. -- Posted by the exoskeleton on 4:14 am on Aug. 8, 2004 i have an idea! why not put wheels on the feet? before you all critisize me think about it. you could have them flip down from the side of the leg and then the task of covering a long relatively smooth distance would be much easyer. i borrowed this idea from the game heavy gear but i dont think you could stick exactly to the game as the have their mech rolling on dirt and sand. this would be a good idea for those mechs that would go into contruction working like forklifts. -- Posted by EdZ on 7:03 am on Aug. 8, 2004 Quote: from the exoskeleton on 1:14 am on Aug. 8, 2004 i have an idea! why not put wheels on the feet? before you all critisize me think about it. you could have them flip down from the side of the leg and then the task of covering a long relatively smooth distance would be much easyer. i borrowed this idea from the game heavy gear but i dont think you could stick exactly to the game as the have their mech rolling on dirt and sand. this would be a good idea for those mechs that would go into contruction working like forklifts. This has ome up a few times before. The usual consensus was that if the mech needs to go long distances, it would be better to carry it on a separatevehicle, and that the extra weight added through the wheels and the power/drive mechanisms outweighs the minimal increase in mobility. -- Posted by Ajay on 10:13 am on Aug. 8, 2004 <>   I'm one of those wheeled mecha fans.  Fitting the drive systems and any kind of suspension into the body is really going to be the biggest problem.  There are a variety of alternatives to a direct drive train though like electric and hydraulic rotary motors that would allow you to place axles almsot anywhere you had a suitable supprt and fluid/electric line access.    Mechaps has mentioned interest in the idea- but not in the Mark-1.  Earle has mentioned that if it is used, then the system would be mounted somewhere on the body, and not in the legs.  This is for stability, and probably makes it easier to mount stronger support systems than you'd be able to in the legs.   The motors and wheels in most of the feet of Heavy gears are just too small to truely be effective.  I do like the look of the Ferret, Pit ull, Grizzly gears, and Naga striders a bit more though.  Gear Krieg has some interesting ideas as well.  These particular vehilces don't have regular legs with wheels tacked onto the feet.  Rather, the lower portion of the legs are more centered around the wheel/tracked system, which can be indepently oriented like normal legs.   Not fast at all when running, but still useful for climbing, straffing, and stepping over obstacles.    The potential usefulness of very small "roller" wheels could be seen when loading and unloading the vehicle.  A 14+ foot tall vehicle is not likely to fit many places in modern society, and there exist relatively few resources already in place to fulfill the role of an adequate transport.  The ability to compact down to a smaller size and still retain some mobility can help when navigating towns, when moving around in warehouses or garages, or when being loaded or unoloaded into a restrictively sized transport.    As far as long range travel is concerned, being totally depenent on a transport vehicle can be somewhat of a handicap.  If possible, I would suggest looking into the application of larger tracks or wheels for higher speed longer range movement.  This might be of more use in a military model than a civilian one though.    Top speed via the legs is could be hampered by the limitations of the balancing system at high speeds, and the wear and tear they would cause on the legs- increasing maintenance and operational costs, while decreasing reliability, range, etc.  It's easy to say "just build the legs strong enough and give it a good balancing" but this is much easier said than done.   To date, there's still no legged vehicle that's ever gone much faster than 5-10 km/h. This is due mostly to limitations in real time balancing programs, not the potential speed and power of the actuators involved. There's ALOT of room for an improvement in speed there.  The powerplant that could theorhetically drive the mark-1 to a top running speed of 150+ km/h would be enough to make a significantly faster tracked/wheeled vehicle which more than likely would not worry as much about maintaining balance with the computer- which Mechaps has mentioned as the most likely limitation of the Mark-1's speed.    Mounting tracks and wheels on the vehicle would also increase maintenance, operational costs, and reduce reliability, but takes the stress off of the potentially more fragile legs.  Whether or the overall outcome is more reliable, less maintenace heavy, and lest costly remains to be seen.    This idea has spawned interest in other legged/wheeled/tracked hybrid vehicles. http://www.mech-eng.curtin.edu.au/HYDROBUG1.pdf http://www.cim.mcgill.ca/~venkat/PUBLICATIONS/JMD_99A.pdf    In regards to the first link and estimates for the performance of their design... Top walking speed- 5 km/h Top rolling speed- 50 km/h   I had another really good pdf file discussing the potential of tracked/wheeled, legged, hybrid, and articulated body locomotion in robots but I can't seem to find it now. (Edited by Ajay at 9:24 am on Aug. 8, 2004) -- Posted by Viking420 on 2:11 pm on Aug. 8, 2004 Wheels on the feet could actualy help with stability. An ankle can only react within a finite range of motion without lifting and repositioning the entire leg. A wheel, either electically or hydraulically driven could reposition the foot without having to lift it. The constant reactionary movements could either be routed to a "toe" contact point with finite travel or to a wheel with infinite travel. The legs would funciton as highly flexible active suspension. A combination of wheel speed/direction constantly updated along with torso position on the horizontal plane could provide a fairly stable platform. It might even be a good place to start with for the application of the balancing system. Picture two computer controlled unicycles with a box on top and you get the basic idea :) -- Posted by the exoskeleton on 2:31 am on Aug. 9, 2004 actually scratch the idea of leg mounted wheels, how about wheels on the back or where its butt would be? then you could just sit it down and press the button and away you go. no need for costly transportation, just sit down and roll -- Posted by EdZ on 6:59 am on Aug. 9, 2004 A more stable wheel platform would be to have the mech kneel down, and place wheels on the lower leg, or on the knees and ankles. Placig them on the lower leg maylose a little stability, but it would be a lot easier to do structurally. -- Posted by Jagarus on 2:00 pm on Aug. 9, 2004 Good ideas and all but remember MPS primary goal is build a working pilotable walking mech, personally i do not see the point in adding wheels to the mech at such an early stage in development. Without going into too much detail, stability will not be a major issue. Well that was more or less no detail. Im still to new to MPS to make sound comments. I just dont remember everything right off the bat. Anyway keep those ideas coming in the meantime. -- Posted by Drachillix on 2:19 pm on Aug. 9, 2004 Quote: from Ajay on 1:00 am on Aug. 8, 2004    Many pilots are MORE than skilled enough to fly aircraft without the aid of computers.  Often times, the computer cushions their actual capabilitiues, keeping them from going to their full potenetial.  Has this doen anything to limit the large scale application of software controlled avionics? Most normal aircraft are self stabilizing due to wing dihedral.  This makes it far simpler to make autopilot routines when the plane can already pretty much fly itself once its in the air. The main difference I can see between the schools of thought on this is proactive vs. reactive balance.  Reactive balancing is far more complex. As far as trick moves like a kick or a punch in a hypothetical combat mech, it would be a simple matter to put marks on the hud for the punck/kick aimpoints and then trigger the "kick script" when your target is in the aimpoint.  The curveball for that is the impact, or as BT players know...the miss. -- Posted by Viking420 on 4:36 pm on Aug. 9, 2004 <> Trick moves? Hypothetical combat mech? I'm no soldier but I kick things all the time, whenever my hands are full and something needs to get moved I kick it or nudge it with my foot. I step on my cigarette butts when I'm done and when a door is jammed I give it a good smack with my boot. Granted a mech won't be smoking, but in a firefighting situation you might want to just stomp out a small "smoker" instead of hosing it down, or a mech loading timber on a truck might need to kick or nudge a piece of rubble out of the way. The flexibility (in terrain mobility and literal limb flexibility) of a mech means that you will probably NEVER have a script for every action you might need to perform, and even if you did then your computer is going to be full of codee that you almost never use. I'm not saying that the operator should have to perform each movement down to every detail, but to limit the controlls to a system that makes improvisation difficult or impossible will hurt the long term performance and acceptance of the mech concept IMO. -- Posted by EdZ on 5:59 pm on Aug. 9, 2004 Quote: from Drachillix on 11:19 am on Aug. 9, 2004 Most normal aircraft are self stabilizing due to wing dihedral.  This makes it far simpler to make autopilot routines when the plane can already pretty much fly itself once its in the air. This is only true for commercial aircraft such as airliners and business aircraft. For sports aircraft, and millitary aircraft, stability is extremely comprimised in exchange for manouverability. The more stable and aircraft is, the harder to make it move in a different direction, and vice versa. These aircraft (at the bottom end of the scale with sports planes) need the stick to be constantly held and monitored (and at the extreme end, e.g. the B2) need a supercomputer to constantly adjust the control surfaces just to keep it in the air. Another parallel is the B2 can fly itself easily, the crew are there in case of an equipment failure, to push the 'big red button', and to assist in take-offs and landings. This is the kind of control computer a mech requires, with the ability to pretty much get about on it's own. Again drawing parallels with the aircraft industry, a construction (or even sports if they get cheap enough) mech would use a less sophisticated computer, but would not give the same degree of accuracy or autonomy. jusrt as the B2 can fly without pilots, and the sports plane needs constant attention, a construction mech would need greater nursing to do it's job. This is no different than any other construction vehicle, and most of the mechs tasks will still be computerised (such as calculating ther pressure needed to grasp and object firmly, the positioning of an arm to place a girder into position, etc) but the overall operation will require a sufficient degree of human overseeing. -- Posted by Drachillix on 6:00 pm on Aug. 9, 2004 Quote: from Viking420 on 1:36 pm on Aug. 9, 2004 I'm no soldier but I kick things all the time, whenever my hands are full and something needs to get moved I kick it or nudge it with my foot. Yes, but you have proactive balance that took years to develop and about 200 actuators contributing in some way to out posture and balance.  When mechs have had as much time to evolve and improve as our bodies I will expect spinning back kicks and ballroom dance.   Our bodies also operate under assumptions of what our speed and direction is.  You can't surprise yourself and make an unexpected move because your body shifts our balance for that move BEFORE WE ACTUALLY MAKE THE MOVE.  Our bodies only react when outside force or poor perception on our part comes up.  So in a nutshell your analogy to our bodies is fallacious.  The mech can't read your mind and adjust balance for a move you havent made yet. -- Posted by Drachillix on 6:45 pm on Aug. 9, 2004 Quote: from Drachillix on 11:19 am on Aug. 9, 2004 Most normal aircraft are self stabilizing due to wing dihedral.  This makes it far simpler to make autopilot routines when the plane can already pretty much fly itself once its in the air. This is only true for commercial aircraft such as airliners and business aircraft. For sports aircraft, and millitary aircraft, stability is extremely comprimised in exchange for manouverability. Thus my use of the terms, most and normal.   -- Posted by Viking420 on 8:18 pm on Aug. 9, 2004 "Proactive balance"? I think I understand what you mean, but I don't entirely agree with you. Besides, the neural net these guys are rigging up fits your analogy fairly well. It learns as it goes, figuring out how to pre-emptively position itself to ensure stability through trial and error (similated I believe to save wear and tear on the actual chassis). Even if the mech isn't sure exactly how to stay up when I tell it to kick an enemy "crunchie" it can still reactively balance itself (awkwardly), every time afterwards it will get better and better at the action thoough. I am of the opinion that a highly invovled interface (but as I said, not overwhelming, finding the balance is the main issue) combined with an adaptive computer and software would make the Mech an incredible flexible piece of machinery. You could evemn, with a capacity for manual controll, order the mech either "pre-trained" or "untrained", depending on your demands for the machine and ability with the system. It would be like dogs somewhat, you can get dogs trained by the breeder or you get puppies and housebreak them yourself. I like puppies :) -- Posted by Viking420 on 8:27 pm on Aug. 9, 2004 And sorry for all the typos... -- Posted by Earle Bishop on 1:45 am on Aug. 10, 2004 Quote: from Viking420 on 5:18 pm on Aug. 9, 2004 You could evemn, with a capacity for manual controll, order the mech either "pre-trained" or "untrained", depending on your demands for the machine and ability with the system. It would be like dogs somewhat, you can get dogs trained by the breeder or you get puppies and housebreak them yourself. I like puppies :) Poor analogy. A better example would be: You can order parts, and build your own computer, byte code a compiler, create a programming language, write, debug and use your own software and operating system... or you can go to dell.com and get the whole package working out of the box with next day delivery. I cannot think if any instance where you would want an untrained mech. You might want it unspecialized for general labor if you panned ot use is on a construction site... but you would still need the machine training and routines for general labor. -- Posted by Ajay on 3:58 am on Aug. 10, 2004  Even if you had a learning computer, I'd imagine that most "training" would be done under very very controlled conditions.  You wouldn't try to use the machine to go through some previeously unencountered maneuver right there on the spot in the field.    It might make more sense to have one learning computer, and several other prebuilt computers that can take the information gained from the other learning computer and already know what the other machine had to risk it's existence doing.   Trying to learn in the field- COULD end up pulling off a new maneuver in a bumbling manner, or you could fall right over, damging the vehicle, yourself, and anyone around you.     <>   Using manual control to any significant degree though would require a good degree of motion capture, which for reasons described earlier would be limited by the operater's lack of sufficient input from outside conditions.    If the detailed control system is do nothing but an additional way of activating scripted movements (lift and move foot forward in kicking motion= initiate kick command), then it is still limited by the computer.     Most exoskeletons use the motion capture and master/slave system with the machine making active adjustments to move with the operator.  Larger vehicles, where the operator is not in as much direct contact with his surroundings as in an exoskeleton, have to make use with a computer controlled system with as many programmed capabilities as possible.   The general motion capture system was applied once to larger vehicles (as far as I know of) to the GE walking truck, but in the complexity of the control system ended up having them scrap the project.  That was in the 60's though. No idea of any attempts at it lately.   -- Posted by Drachillix on 6:19 pm on Aug. 10, 2004 Quote: from Viking420 on 5:18 pm on Aug. 9, 2004 "Proactive balance"? I think I understand what you mean, but I don't entirely agree with you. You can disagree if you like, but thats how our bodies do it.  You are asking for a "psychic" mech.  Balance compensation would have to be executed before the move begins. Besides, the neural net these guys are rigging up fits your analogy fairly well. It learns as it goes, figuring out how to pre-emptively position itself to ensure stability through trial and error (similated I believe to save wear and tear on the actual chassis). Even if the mech isn't sure exactly how to stay up when I tell it to kick an enemy "crunchie" it can still reactively balance itself (awkwardly), every time afterwards it will get better and better at the action thoough. Nope because it doesen't know what you intend to do until you do it.  People don't make precisely the same moves from the same point in stride every time.  In addition our bodies adapt to different activities.  Muscles strengthen or atrophy based on the type and frequency of motion we perform.  There are physical changes that occur in our bodies in response to activity.  Mechs won't grow bigger engines or stronger hydraulics as they work. I am of the opinion that a highly invovled interface (but as I said, not overwhelming, finding the balance is the main issue) combined with an adaptive computer and software would make the Mech an incredible flexible piece of machinery. Ah, yes.  We will utilize symbiotic synergy to create a paradigm shift along a new vector better facilitating our core competencies. Or we could just use a neurohelmet.... -- Posted by EdZ on 6:34 pm on Aug. 10, 2004 Quote: from Drachillix on 3:19 pm on Aug. 10, 2004 or we could just use a neurohelmet.... Or use a mech with a computer powerful enough to calculate the leg and foot positions needed as soo nas you push the pedal. Oh wait... -- Posted by avantgarde05 on 6:37 pm on Aug. 10, 2004 i think that the best approach is to use a computer system that ientifies a certain object a can interact with it.because the basic principle is to get the mech to be able to manuever around on it's own and be able to perform a specific task. It really depends on what the role of the mech is desgined for. And by the looks of the mark series it sort of looks like it would be best suited for construction. It would be a big help to have a mech around to finish building your little mech factory. The control stick system i think is a really good idea. mocap sounds horrible (what the mech trips over a frickin pebble or something and falls) i woul feel really sorry for the guy inside the mech. but the gundam idea doesn't sound bad. If i were the designer i would program simple object like a steel i beam into the system, be able to target it and interact with it after wards. It sounds really simple but i know it's hard to get it really on the mech but hey, jus an idea. Just watch some mech oriented movies and you'll see. -- Posted by Viking420 on 8:35 pm on Aug. 10, 2004 All I'm really saying is that an extra dimension of controll with the feet brings in another area of controlls (for whatever function) and (depending on execution) does so in a fairly ergonomic way while taking up minimal space. I think that with another controll for the torso orientation (or whatever your using as an inverted pendulum) and feedback all around you would have an additional way of manually programming new movements (wich the neural net would refine). Of course you wouldn't just make up new moves "in the field" under most circumstances. It would be like a police officer trying a new shortcut on the way to a crime scene... do what you KNOW works when it matters, but if you've got a few free moments then go ahead and try something new because it might pay off. There are several potential arguments against this idea that I can think of personally. 1. You plan on selling proprietary movement "scripts" and don't WANT people creating them for themselves. 2. Humorless SOB foremen/sargeants/etc don't ever want to see a mech skipping down the road... 3. There IS potential for accidents, I'd be lying if that wasn't a worry of mine too. The fact is though that people will ALWAYS find a way to injure themselves or others, no matter how safe you try to make things. Do your best for sure, but don't limit yourself because of the fear either. If we only did safe things nobody would have left the trees and we'd still be flinging crap at each other :) -- Posted by Earle Bishop on 9:15 pm on Aug. 10, 2004 Quote: from avantgarde05 on 3:37 pm on Aug. 10, 2004 i think that the best approach is to use a computer system that ientifies a certain object a can interact with it. This is more or less what the Bi-optic Imaging System does. The BIS consists of 4 cameras, and does edge detects and frame compares to identify objects in its field of interaction. -- Posted by Ajay on 12:53 am on Aug. 11, 2004 <>   But where exactly is the machine to try out new maneuvers without the threat of falling down?  A failed maneuver could very equal smashed up machine/operater.  Wherever you train needs to have facilities t prevent something from getting busted when something goes wrong.     A police officer can always consult a map ahead of time to help hedge their bets when finding a new shortcut, and even if the shortcut turns out to be bust, they can just back up and turn around.  A failed maneuver could very equal smashed up machine/operater.     A better anology would be officers training for chases, where they are requied to maneuver through cones and around obstacles to get a better feel for their vehicle.  This is of course preceded by class room instruction, and the actual tests are carried out in a much more controled enviroment.  Cars and other wheeled vehicles have a passive stability you're not going to find in a legged vehicle that make them much less prone to falling over due to unpredictable/sharp turns and sudden maneuvers. <<1. You plan on selling proprietary movement "scripts" and don't WANT people creating them for themselves.>>    Not unless they have the proper facilities to do so.  If the vehicle already has a very large number of movements and is constantly comming out with new ones, then this shouldn't be too much of a problem. <<3. There IS potential for accidents, I'd be lying if that wasn't a worry of mine too. The fact is though that people will ALWAYS find a way to injure themselves or others, no matter how safe you try to make things. Do your best for sure, but don't limit yourself because of the fear either. If we only did safe things nobody would have left the trees and we'd still be flinging crap at each other >>   Doing things which are obvious safety hazards don't help though.  If you're going to be dangerous, there are plenty of thigs you can do to increase your chances of not getting hurt or doing damage.  In this case, damage is not just superficial- it will cost money and time repairing the machine and possibly the operator in the event of a failure.    Perhaps this is a situation where developer would be able to contribute more ahead of time than the test pilot. <>   It's a new deimension in the legs, but it really doesn't add and extra "dimension" of control.  It's already been pointed out that space for vertical leg movement will already be at a premium, leaing you with too little for effective motion capture, especially for controlling actions like kicking and jumping.  These might also necesistate the use of arms for maintaining balance, and unless you intend to have operater using joysticks or motion capture to orient the vehicle's upper toroso and limbs in order to miantain stability while it's going prone, or performing a jump or kick, there's still going to be some aspects of control that are very much scripted- perhaps with the entire machine acing totally autonomously on a command for a few seconds.     This practically necesitates scripted movement, and reduces foot controls to just a way of initialising vehicle leg movement.  2d controls, with a pair of buttons on the joysticks/somewhere else in the cockpit to control more abstracted elevation and jumping/kicking in the rare events they are used don't take up much more room, and probably take up less considering that you also don't have to install the more intense display systems and possible force feedback system that would make the lower legs master/slave system feasible in the first place.   If done,  I can say it would be somewhat more ergonomic- slaving all leg related motions to the foot controls- but I don't say that the ideal overall is as practical as the 2d foot control system.     No the current idea floating around doesn't allow for user customization in the field, but under most circumstance, the original number of actions, in addition to updates through testing in a more controlled system should be more than enough to keep the operator from possibly damaging his machine and himself trying to develop totally new ones by themselves through actual operation of the vehicle. -- Posted by Viking420 on 11:09 am on Aug. 11, 2004 Its just an idea to enhance flexibility in the field under uncertain conditions. If your worried about a spontaneous move causing loss of controll then are you less worried about the mech losing controll when encountering unfamiliar terrain or bad footing? I already have stated as well that a torso orientation controll would be virtualy mandatory with a manual balancing system. I understand that a manual balancing and controll system would not be practical for day to day operation and so I am presenting the idea as a means of training the NNet rather than a primary controll method. The details need a lot of work, I've always tried to be careful to make that point, but I believe the potential benefits are worth it. Besides, a mech that can't tale a fall isn't going to make it to market, and any prototype would hopefulyy have "training wheels" or something to protect it while you iron out the weaknesses. I can see this sort of "symbiotic" controll system being held back until the later generations of mechs for sure, it WOULD likely be too much to deal with in the Mark 1 stage of development. All I'm suggesting is that you not completely discount the idea in the long term. -- Posted by Ajay on 11:58 am on Aug. 11, 2004 <>   Yes I am; especially terrain that can shift under it like gravel, sand, etc.  I'm not the only one worried about the instability of bipedal walking vehicles either.  Up until Mechaps, I can't think of anyone who has even attempted it. <>  I personally think they would, granted they aren't totalled when they do so, and falling doesn't happen too often to begin with.  it might take more than one would think to get a of tons of interconnected actuators, computers, and other pieces to hit the ground from a height of several meters and expect nothing to get jumbled around.   However, as I'm not aware of the usual number of steps per fall most bipedal robots these days are capable of, my expectations might be off. << I already have stated as well that a torso orientation controll would be virtualy mandatory with a manual balancing system.>>   Who's controlling the torso though and to what degree?  Are you going to make the operater responsible for both the orientation of the legs and the torso?  Is the operater moving the legs, and the torso making automatic adjustments?   The following may have some point, but I will admit before hand that it is already presented from a very biased viewpoint... - but if anything should be directly controlled by the perater, it should be everything above the waist.  When you give the vehicle control of the upper torso, you take away some of the control of the arms, and by extension the manipulators, and these are ultimately the tools through which the human element trully comes into play.   Although the vehicle's legs can get it places, the arms/manipulators are what really what allows it do anything, whether it be picking up objects, or placing them in different spots.  This is where the decision making of the operater is most important in comparison to the capabilit of the computer, so the operater should be ale to perform this task with less interference.   Now that that's done. ..................................................... <>   Which is why I suggested the possibility of a single NNet being able to share it's experiences with others.      In this case, the original computer and host vehicle could make use of a control system that would be totally impractical for general operation- possibly including full body motion capture or individual joint placement with a bunch of people all standing around looking at a console (common in modern robotics).   <>  Duly noted.   I suppose that right now, I'm being conservative (which is what I tend to do).  When things are working and working quite well- like heavier automation in larger walking machines- I tend to be less inviting of relatively new ideas, especially those that I have understood to be implemented in the past with less than ideal results- like the manual control scheme with a computer actively adjusting to keep the legs oriented correctly.  I tend not to count anything out forever, as I understand many situations can change over time.  However, until that time actually comes, I believe that previous experience should account for some fair degree of skepticism.  Although I may disagree with you now, I can garauntee you that if somewhere down the line, your proposed idea is pulled off succesfully, I will gladly acknowledge it and perhaps become a vehement supporter. -- Posted by Viking420 on 12:48 pm on Aug. 11, 2004 <<  I'm not the only one worried about the instability of bipedal walking vehicles either.  Up until Mechaps, I can't think of anyone who has even attempted it.>> Check out the Springwalker, I believe it has been clocked at 35+ kph with completely manual balancing AND a human powerplant ("pedal" powered). One of the aspects of my interface (wich I have been working on for several years) is the ability to instinctively deal with a fall. I've got a few years of Judo and rugby under my belt so I KNOW how to fall (and be thrown) without injury and I have tried to incorporate some of those techniques into my proposed controll interface. With a torso controll held in the hands (flight yoke style) a forward fall would be instinctively reacted to by pushing your hands forward, along with thumbstick (or a similar system) controlled arm actuation it would take relatiovely little training to get a pilot to put his mech's hands out to catch himself while orienting the torso forward to lessen the overall impact. Combined with shocks in the major load bearing joints this could help prevent major damage from falls. Of course even healthy humans sometimes break bones from falling... gravity is a bitch :) -- Posted by Ajay on 1:26 pm on Aug. 11, 2004  Springwalker's an exoskeleton though.  It still gets the benefits of the extra large field of view and greater degree of operater contact with the ground, and operater controlled balance I mentioned earlier.  I'm more interested in seeing walking vehicles- not exoskeletons- using a more manual control interface.     I can understad operater control of the arms for bracing for a landing, but I'm more interested in torso balance and arm control during more radical movements of the legs. (Edited by Ajay at 12:28 pm on Aug. 11, 2004) -- Posted by Viking420 on 5:21 pm on Aug. 11, 2004 A stabalized cockpit combined with a controll yoke slaved tot he actual position of the torso relative to the waist, or a reverse of the situation (stabalized controll in a static cockpit) is the method I have in mind. As the most dextrous appendages, I feel the hands are best suited to the primary balancing system (torso as an inverted pendulum). With direct manual controll you would "walk" as much with your hands as you do with your feet, maybe more so in some situations. You lean forward and to the right when stepping with the right foot, forward and left for the other, etc. This controll would probably require some form of feedback to operate effectively, as well as calibration for each operator. My idea of a mech really is very close to an exoskeleton, the difference is primarily one of proportion as I see it. I also WOULD call an exoskeleton that can enhance physical performance a vehicle to a degree, the line between these concepts is very blurry at this stage. Also, with feedback controlls and a cockpit designed to give a good view of the surroundings there is no reason a mech couldn't be operated in a similar way to an exoskeleton. The controlls would not be as complex and involved as a form fitting exoskeleton for space and safety reasons but with computer assistance there is less need. Of course, I also recognize that the mark 1 is a concept prototype and that the main idea is to test the computer and balancing systems on a large machine, nothing more. I can see trying a more involved interface on the mark 2 or 3 though, once confidence in the underlying systems is achieved. -- Posted by Ajay on 3:25 am on Aug. 12, 2004   Hmmm well this seems to be getting quite complicated.  Rather than say "it's goign to be tougher than you'd hink", I'll let the actual engineering and development process show us just how diffuclt the "motion capture without really being motion capture" style of control can be in a large walking vehicle. -- Posted by Earle Bishop on 5:15 am on Aug. 12, 2004 <> This was my feeling as well. He has not actually tested any of this yet, it is all pure conjecture; unproven hypothesis. -- Posted by Drachillix on 2:04 pm on Aug. 12, 2004 Quote: from Viking420 on 5:35 pm on Aug. 10, 2004 There are several potential arguments against this idea that I can think of personally. 1. You plan on selling proprietary movement "scripts" and don't WANT people creating them for themselves. Especially when the fall on their ass, try and blame it on the mech and or balancing software and sue MPS for damages to the mech and pilot. 2. Humorless SOB foremen/sargeants/etc don't ever want to see a mech skipping down the road... They hate forklift jousting too.  So if you are skipping along and fall is it the mechs fault or did your use of a non standard gait pattern contribute to the fall.... 3. There IS potential for accidents, I'd be lying if that wasn't a worry of mine too. The fact is though that people will ALWAYS find a way to injure themselves or others, no matter how safe you try to make things. Use of movement scripts not made buy MPS of created by a "licenced MPS movement script solution provider" would have to be specified as some kind of unauthorized use in purchase agreements to avoid the liability nightmares. The boogeyman you need to be concerned with is not the foreman, its the plaintiffs attorney. -- Posted by cakefool on 2:10 pm on Aug. 12, 2004 Warning - teaching this vehicle to dance may invalidate your warranty - however if you do manage to teach it to tango, please get in touch with recruitment on 555... -- Posted by Viking420 on 5:57 pm on Aug. 12, 2004 Yes, Earle, the concepts I am presenting to the Forum are untested theories. Did you really expect more than that from me? I have worked very hard on my ideas, and considering the extremely limited resources at my disposal I think I've done fairly well with my little hobby of designing an interface to operate a bipedal vehicle. I have indeed taken a different approach to my strategy than you have, but since neither of us has a mech in our garage I'd say neither of us has hit the mark yet. And why are you worrying about potential lawsuites from injured operators at this stage? The Wright brothers weren't thinking about the Enola Gay when they built Flyer 1, they were focused on proving powered flight was achievable. Flyer one flew with one of the Wrights (I forget wich one) pulling on the wings and leaning for controlls... it eventualy evolved into digital fly by wire controlls controlled by a joystick and throttle in the most advanced modern aircraft. -- Posted by EdZ on 5:59 pm on Aug. 12, 2004 The motion capture-feedback system might not be as useful for teaching applications as you think. If you are trying to create entirely new 'moves' for a mech, and don't want to cause damage to whatever mech you are testing on (or the minimal of damage), then it would probably be best to model it entirely on computer, and use an evulotionary model to test all the different combinations of joint movements and weight distribution to find the stablest (or fastest, or strongest, depending on the application) move possible. This is not possible with the mo-cap system as you are thethered to using moves that  a human poerator can perform, which most likely will not be the optimal series of motions for the mech itself. -- Posted by Viking420 on 9:45 pm on Aug. 12, 2004 Alright, any physical controll is technically a motion capture, there are differences of scale between a button or lever and a full body waldo but they all capture movement to use as input. All my idea involves is position within a cone for each leg instead of just position within a circle. A hatswitch controll for each foot and a joystick for each hand is 8 total linkages. All I'm suggesting is adding 2 more linkages and adding feedback to them. The vertical foot controlls could be operated by your toe and the feedback is only there to inform the operator of ground contact, simple vibration would work as the only point is to let the operator know when each foot is planted and when they are not. The feedback I would prefer for all controlls would be variable shock absorbers, though. The computer might not be able to react to some inputs its true, that is where the shocks come in, as the position of a foot nears these limits the shocks stiffen, as the feet return to a more stable position they soften. The details of how a mech reacts and keeps balance with it's feet in any given position relative to its body is up to the computer, as is controll over these shocks (6 total). Thats 6 potentiometers and 6 variable valves, hardly the most daunting system involved in a mech. -- Posted by Drachillix on 10:46 pm on Aug. 12, 2004 Quote: from Viking420 on 2:57 pm on Aug. 12, 2004 I have indeed taken a different approach to my strategy than you have, but since neither of us has a mech in our garage I'd say neither of us has hit the mark yet. Neither does Earl Bishop, but I sure envy his garage :) And why are you worrying about potential lawsuites from injured operators at this stage? Ignoring legal reality is a sure fire way to find someone else owning the worlds first mech factory. The Wright brothers weren't thinking about the Enola Gay when they built Flyer 1, they were focused on proving powered flight was achievable. Flyer one flew with one of the Wrights (I forget wich one) pulling on the wings and leaning for controlls... it eventualy evolved into digital fly by wire controlls controlled by a joystick and throttle in the most advanced modern aircraft. Interestingly enough your suggestion has more of the "lean" methods while everyone else is working more along the fly by wire method. -- Posted by Ajay on 4:03 am on Aug. 13, 2004 <>   Well one proposed control system (fly-by-wire through 2D controls) has already established itself as the basis of most all walking vehicle leg controls.     This must lend some credence to it's capabilities.    Again, the last attempt at more detailed control of a larger vehicle using more detailed input from the operater's feet and legs didn;t turn out so well. <>    Physical control is usually called analogue.  Motion capture is generally when the movement of the device mimicks the operator's at a recognizable ratio.    Most joysticks and hatswitch controls would be considered analgue to some degree, but most definately not motion capture.  You proposed system is not trully motion capture either, and is simply a more involved way of delivering many of the same general digital commands. <>  As well as the space involved, and a possible readjustment of the operater's seat.   <>   A point that many have been trying to make so far is that succesful pidal walking requires more than "reaction" to the ground once you're standing.  Placing the foot solidly on the ground means orienting it to some degree before it even hits it.      You can initiate leg movement, but even then, the exact place where the foot hits the ground is still decided by the vehicle.   I still really can't see what you can do with this system that you can't do with hatswitch controls.  You assert that there are maneuvers that can be perfomed with this control style that cannot be done with hatswitch controls, but many have come up with suggestions as how these particualr maneuvers are not only able to be carried out through 3-xis foot controls.    You are, however, increasing the ammount of operator attention to the process of walking and maintaining balance.    I'm more inclined to making the walking process simple, and thereby creating a stable upper torso, more capable of operating without effecting the lower one, and as such, allowing the operator to focus more on maneuvering the arms, manipulators, or any other systems while performing their job.   -- Posted by Earle Bishop on 7:23 am on Aug. 13, 2004 Quote: from Viking420 on 2:57 pm on Aug. 12, 2004 Yes, Earle, the concepts I am presenting to the Forum are untested theories. Did you really expect more than that from me? I have worked very hard on my ideas, and considering the extremely limited resources at my disposal I think I've done fairly well with my little hobby of designing an interface to operate a bipedal vehicle. I have indeed taken a different approach to my strategy than you have, but since neither of us has a mech in our garage I'd say neither of us has hit the mark yet. Neither of us have mechs. I, however, have mech parts. What you fail to realize is that MPS has processed THOUSANDS if not tens of thousands, of good ideas. We debated, brainstormed, researched, hypothesized, wrote code samples, built and tested models, circuts, etc. What we are left with are not good ideas, but good ideas that have been proven to work. Edison made hundreds of attempts at lightbulbs. Only one really worked. Both of us have lightbulbs. Yours are not yet proven to light. And yes, I do expect more from you. I expect you to spend as much time making models and testing stuff or talking to people who can do the same, as much time as you spend on this board trying to sell a broken concept that we continually spend the time to show you is at fault. -- Posted by Viking420 on 12:37 pm on Aug. 13, 2004 <> Well, I guess I'm flattered, Earle, but still frustrated. I make on average $5000 a year, that's Canadian money and I'm not forgetting any zeroes. Over 75% of my income goes directly into basic housing, and most of the rest goes towards food. I can't even afford new shoes or a bike. I am looking for a better job, I am fixing up my home, I am trying to stay involved in local politics and I am doing my best to self-educate myself in a fairly well rounded manner. I'm genuinely sorry I can't bring more to the table, but its all I've got. Frankly I don't think my concept is "broken" as you put it, it is at the extremely early stages of development and needs many iterations of refinement before it can become either a working system OR a failed prototype. Of course my spiral notebook looks amateurish next to the MCF, it's like an abacus next to Deep Blue. I've said before that my own design is a far different overall concept than the Mark 1 anyways... I don't know if it will work (I actualy doubt it myself) but even a failed experiment can result in new knowledge and ideas and that is all I'm trying to share. -- Posted by Viking420 on 12:00 am on Aug. 14, 2004 If your not interested, I can accept that. Problems keep being discussed where I feel aspects of my system would help... the latest one was a discussion over a dead man's failsafe system and my point was simply that repeated movements rather than a maintained position as forms of operator input would inherently function as a sort of dead man's system. Of course, that simple suggestion seems to have gotten lost in the waves of broadsides launched over completely unrelated weaknesses I have already admitted to in my concept. -- Posted by EdZ on 5:36 am on Aug. 14, 2004 On that point specifically, if a man became unconscious when operating your system, his legs would crumple up and he would fall over. So would the mech, causing damage to anyone/anything around it. Continued movement isn't the only nor the bigest hazard with an operator emergency or a malfunction. -- Posted by Viking420 on 1:10 pm on Aug. 14, 2004 First off, an unconciuous operator in my cockpit layout would not "crumple" as you so put it, he's seated on a bicycle or motorcycle saddle and secured to the mech at the shoulders, hips and knees. Anyways, my cockpit idea is NOT going to be used, I accepted that a long time ago but I still believe there some concepts from my plan that could be adapted to virtualy any other interface. My operator initiated footsteps idea WOULD remove direct computer controll over one aspect of balance and movement, stride frequency. THAT is the legitimate weakness (or strength?) of my idea, and I admit it completely as it is in fact the point of my idea. Any other real or percieved problem is in fact a critique of the specific methods of accomplishing this very simple idea of controll, not a critique of the idea itself. THATS what I'm getting frustrated about, everyone is dismissing my ideas because of HOW I propose to implement them rather than discussing the usefullness (if any) of the overall concepts. -- Posted by EdZ on 3:07 pm on Aug. 14, 2004 Seeing as the only way for the mech to balance is for the computer to calculate stride langth and foot placement, it woudl not work for normal operation (as you said). IF you needed a specific movement to be scripted, then you would use a computer to calculate optimum foot placement as the human body would not be optimised for the mech's structure (I'm sure I said this before). -- Posted by Viking420 on 6:21 pm on Aug. 14, 2004 Your not paying attention, thats yet another potential point you raised against my specific cockpit idea, and its not even related to the general controll idea I am presenting. Foot placement and stride length are not directly affected at all by my idea, just stride FREQUENCY. Not only are you "kicking a dead horse", your kicking the wrong dead horse. -- Posted by Gideon on 1:40 am on Aug. 15, 2004 realistically, the best we can do is a joystick and throttle using a computer program that anticipates what the pilot is attempting. If weapons on it are armed, for example, responses should be sharper. If weapons are armed but the pilot is moving slow, the computer should allow for slower, more silent responses. As for hands, they are too complicated and, to be perfectly honest, a rather stupid and useless feature. Hand to hand combat has been completely erradicated from today's wars, and all it would take for the machine to kill a man is hitting him with the barrel of his weapon. For practical use (such as opening smashed up cars to rescue the people) can be done using a saw and arm with two "fingers" (such as the loader from Aliens) that are controlled by smaller sticks allowing for 360 degrees of rotation, mounted either straight to the control panel, or on the stick and throttle. The pilot should be able to alternate between them, allowing for up and down, and left and right movements with either stick. There are computers such as this in several hospitals that allow doctors to perform complex operations from another room and, theoretically, from across the world. Most weapons should be mounted and controlled by the same system used by apache and comanche helicopters. The pilot simply looks in the direction he wants the gun to point. This would allow for very accurate shoulder mounted weaponry. A second gunner/navigator would also be very usefull in targeting. Systems from the M1A2 could be utilized to allow for target aquisition and tracking despite vibrations and the natural rise and fall created by bipedal motion. Essentially, most of the machine would be automatically controlled by a computer or computers. the legs would have to be computer controlled. I know this takes a lot of fantasy out of the big robots, but its reality. Gundam, Macross, and gasaraki are simply fantasy. Pure and simple. Gozilla will come to life long before we'll ever build something as phony as a gundam. -- Posted by Earle Bishop on 2:22 am on Aug. 15, 2004 Mind you, most of these arguments have been raised in the past, so I have am prepared response for most of this. Quote: from Gideon on 10:40 pm on Aug. 14, 2004 realistically, the best we can do is a joystick and throttle using a computer program that anticipates what the pilot is attempting. *chuckle* and how in the fsk do you propose to have the computer guess what you are trying to get it to do with something you are holding. You forget that this is not a car (which would be silly to control with a joystick and throttle) or a single directional plane. It is a land bound machine, with no tolerances for banking or diving. It must be capable of steering around not buildings hundreds of feet apart, but alleyways and levels of a construction site. Planes and cars lack arms. As such, I cannot see how one would possibly control with a simple joystick and throttle. Both items are in our inventory, and will be installed on the machine, but our joystick combo is overkill. It has about 5 joysticks worth of features and buttons onboard. Although JS and throttle are handy, we will still be using mo cap for hand movements, and retinal position detection for area specific targeting. As for hands, they are too complicated and, to be perfectly honest, a rather stupid and useless feature. Hand to hand or melee combat is straight out not going to happen. Most of us agree on that. You forget that hands allow a machine to work like a human. It can pick up mech-sized shovels and dig trenches; it can make and move mech sized sandbags, or carry prefab structures and assemble them. You can recover a dropped weapon from a downed machine in combat. There is nothing inescapably complex about hands, in manufacturing, control or use. All of these elements have been done to death, long before MPS came to the table. A second gunner/navigator would also be very usefull in targeting. Look for us stuffing more people into later versions of the design. For now, we barely fit the small test pilots we have. despite vibrations and the natural rise and fall created by bipedal motion. The pilot will be padded from quite a lot of this. Essentially, most of the machine would be automatically controlled by a computer or computers. the legs would have to be computer controlled. This is the basic idea of how our system operates. A pilot does have the capability to place feet wherever he likes, but by and large it is a 100% machine task. I know this takes a lot of fantasy out of the big robots, but its reality. Gundam, Macross, and gasaraki are simply fantasy. Pure and simple. Gozilla will come to life long before we'll ever build something as phony as a gundam. You said the "G" word. I now have to go find and burn some Gundam Wing DVDs. Do not worry about robbing us of gundam aspirations. Everyone working on this project is a realist. We are here for the long haul.We choose components that are easy to acquire, maintain and repair, in every section of the machine. You seem to have strong opinions on what you believe will create an ideal interface. Unfortunately, you have not shared much as to your reasoning behind each of these assertions. Aside from the stuff I directly agree with from five years of work and research, I am interested to hear the roots and conception behind the rest of your ideas. For the most part, you seem to have hit the proverbial nail on the head. -- Posted by Gideon on 2:43 am on Aug. 15, 2004 I wasn't really trying to push off the engineers, but rather trying to get people to see that 40 foot robots that shoot giant laser beams from their eyes aren't exactly within our grasps. Its rather late right now, and I'm off on vacation for a week starting early tomorrow, but I'll try to answer what I can. <<*chuckle* and how in the fsk do you propose to have the computer guess what you are trying to get it to do with something you are holding. You forget that this is not a car (which would be silly to control with a joystick and throttle) or a single directional plane.>> I haven't forgotten. I know all too well. But I have seen systems that work as I outlined in my original post. I'm not a programmer, and as such I can't tell you exactly what to do, but you have to write it so the computer reads a certain series of inputs in relation to pilot control. For example: weapons are armed, pilot is moving slow, torso is scanning from one side to the next, therefore: reduce stride to low the chance of being tripped and reduce response time, lower gearing to prevent wear and excessive noise, and the choices can go on depending on what you do. This would eliminate many of the buttons needed for doing something humans do automatically. <> again, I have not forgotten. All that you've mentioned is fine and dandy, but have you designed fingers and actuaters strong enough to lift a prefabricated structure? No. Would sandbags actually need to be that big? no. Would the machine ever get close enough to lift the weapon off another? not likely. even within the group, there isn't enough time in combat to stop and recover the weapon. Its better to take a large ammo hopper and belt feed it into a weapon thats bolted the the arm. You also have the problems of accuracy, reloading (where exactly would the robot store extra magazines?) and other such things. For non-combat uses, it could utilize more of a forked hand as apposed to a human hand. I know your well versed in what NASA has already created as far as robotic arms are concerned. <> I'd help you do that. I can't stand the series. If your curious as to why I have these ideas, its because of the exposure I've had to the actual military (no, not the fictional Japanese army in anime's). My entire family has a history of fighting in wars, and my best friend is currently in Iraq. My father was a gunsmith before and after he was a sniper in vietnam, and my brother is still a Captain in the army, after serving in Desert Storm. I've spent a lot of time (probably too much) looking at military equipment and firearms in general. And as the saying goes, "If it ain't broke, don't fix it." -- Posted by Earle Bishop on 3:34 am on Aug. 15, 2004 Quote: from Gideon on 11:43 pm on Aug. 14, 2004 But I have seen systems that work as I outlined in my original post. What systems, specifically? I am very curious. I'm not a programmer, and as such I can't tell you exactly what to do, but you have to write it so the computer reads a certain series of inputs in relation to pilot control. I am a programmer (amongst other things) and making a system mode switch that specifically seems a bit like folly. A mech is not exactly capable (at least MPS' incarnation) of sneaking up on an opponent. I suppose that is the kind of thing that gets proven or disproved in a field test. We have all manner of neural networks in tow, so the system is far from incomplex. The biggest setback would be defining these 'modes' of operation. This is why there are mostly ex military and gamers designing our interfaces. have you designed fingers and actuaters strong enough to lift a prefabricated structure? No. You are quite correct. We never bothered designing arms or hands, because it would cost us less to purchase a pre-designed system that trying to manufacture our own. We never engineered one, but went on the basis of existing paired articulated robotic arm systems some 4 years back when we were first trying to figure out weights. Would sandbags actually need to be that big? Why not? Would the machine ever get close enough to lift the weapon off another? not likely. From a downed opponent or ally, yes. Quite easily. even within the group, there isn't enough time in combat to stop and recover the weapon. Having never been in combat, I cannot say. Having been in combat like situations, stealing ammo from a 'dead' ally made our team win that particular battle. Also, I do recall reading about a marine in GW2 that picked up and emptied several weapons in the pursuit of killing something like 29 iraqi soldiers. Its better to take a large ammo hopper and belt feed it into a weapon thats bolted the the arm. Once your arm is damaged, you can no longer fire the weapon (unless you have hands, and switch it out). You also have the problems of accuracy, Explain how this would cause an accuracy loss. reloading (where exactly would the robot store extra magazines?) and other such things. Probably the same place a human does. I certainly do not want unmagazined ammo jumping around in a hopper getting jammed. For non-combat uses, it could utilize more of a forked hand as apposed to a human hand. I know your well versed in what NASA has already created as far as robotic arms are concerned. Yes, none of it is very much for the civilian market. On a construction site, you have a bulldozer, which can be fitted with a bucket. If that machine breaks, you either have to get another, or do without until it gets repaired. A broken mech can pass its digging equipment to another machine. I will not delve into the reusability of simple tools from one machine to the next. Instead of having one welding rig, you have any number of machines capable of executing the same task. Hands (at least, in some incarnation) are a foregone conclusion of necessity. I will let the other board members contest this point. If your curious as to why I have these ideas, its because of the exposure I've had to the actual military (no, not the fictional Japanese army in anime's). My entire family has a history of fighting in wars, and my best friend is currently in Iraq. My father was a gunsmith before and after he was a sniper in vietnam, and my brother is still a Captain in the army, after serving in Desert Storm. I've spent a lot of time (probably too much) looking at military equipment and firearms in general. I was curious as to your specific reasoning, more than anything. A lot of your background is pretty evenly matched. My father was a cold warrior in the 80s, and I actually went on base at Sandia Nat’l labs more than once (where my father worked and I lived in Albuquerque for 6 or 7 years). I lived in Toole, Utah, and went on base (where my father worked) where over 90% of the US’s chemical weapons are stored. I never visited the Savanah River Site where my father worked on several occasions, but I was exposed to the stuff he was allowed to talk about. I will one day own a Unimog. Just beyond the reach of my computer mouse are more than a dozen spent .357 casings. I presently live 7 miles from Edwards Air Force Base, and regularly shop in stores with people in full military fatigues. 70% of our toolboxes are ammocases. I am currently wearing a US Army issue shirt stolen from one of my enlisted ex’s, and olive drab cutoffs. One of our interns is presently being deployed to Iraq. My father also went to gunsmithing school, and I underwent sniper training, as well as all sorts of arms handling. Almost everything is legal in Utah  =) That is just my story. The rest of the team outshines me by far. Dickerson has over a decade of military, and our actual test pilot has something like 1400+ hours of flight time, 250+ carrier landings, and a flight test history that covers everything that was ever done to the F18. And that is only part of the stuff he has disclosed thus far. I was interested in not the history of your ideological influences, as I hope I outlined above, we would probably see eye to eye on that. I am interested in the specific details that have you decide to choose one option over another. More of the 2+2 and less of the equals 4. Nothing is truly broken with wheeled or tracked vehicles. I simply intend to fix things better than they presently are. The last job I quit to do MPS full time was doing facilitations management for a Japanese company in Los Angeles. By the time I left, each person was doing ten times the work they were capable of 6 months before my system was deployed. This is not conjecture on my part, but a statistically proven item that I documented and exposed. Nothing was broken. I simply found a better way to do it. -- Posted by EdZ on 5:21 am on Aug. 15, 2004 Quote: from Viking420 on 3:21 pm on Aug. 14, 2004 Your not paying attention, thats yet another potential point you raised against my specific cockpit idea, and its not even related to the general controll idea I am presenting. Foot placement and stride length are not directly affected at all by my idea, just stride FREQUENCY. Not only are you "kicking a dead horse", your kicking the wrong dead horse. If you want to control stride frequency, place a scroll wheel on the joystick and use that. And in what applications would it be advantageous to have precise control of the stride frequency anyway? -- Posted by ChrisDickerson on 1:07 pm on Aug. 15, 2004 Pay close attention to how you negotiate hills, ice, water and flat sufaces. You automatically adopt a different stride frequency. This enables you to move while maintaining your balance. Too long a stride on ice and you and the ice get to know each other a bit better.. See what he means now? -- Posted by Gideon on 1:18 pm on Aug. 15, 2004 Was going to type out my explanation, but I have to leave for vacation now. I promise I'll pull you up to speed when I get back though. -- Posted by Archanasia on 2:13 pm on Aug. 15, 2004    Having no programming experiance whatsoever I personally would avoid having to get a computer tied to the control systems at all costs.  It seems like it would be nearly impossible for a mech to navigate anything but a flat surface without tripping and falling.  However, if you can get a joystick/petals/whatever to work in uneven terrain with a computer program then thats fine, it just seems harder to do.    Personally, I think the best option (the simplist) for control would be some sort of body suit that monitored the joint angles in your body and then used a ratio to translate that into movement of the hydrolics.  The pilot would have to be suspended in midair and not touching the floor.  There would be a learning curve as the pilot would have to learn how move his limbs in such a way as to move the mech, which would have a diffrent stride, etc. as a human.  This method is not too hard except for the hands, where you would have to deal on a much more detailed and smaller scale system of sensors.  The sensor hand thingy that I saw earlier on the site seemed to me like it was basing finger position on the overall bend of the fingers and not actual joint movements so I would be suspicious of how precise of a hand movement you could acheive with something like that.  That's just me guessing though 'cause I haven't bought one yet.  :) -Just to clarify the "body suit" concept.  I didn't want to suggest the idea of some nice slimfitting futurist thing that will never exist.  I was thinking of something more like the braces you might see on someone when they have a bad knee or break a bone near a joint.  Basically a metal joint placed at each joint in order to monitor the angles.  Even for complex joints such as the shoulder, it would still be a rather simple design and wouldn't affect pilot movement. -- Posted by EdZ on 7:23 pm on Aug. 15, 2004 Quote: from ChrisDickerson on 10:07 am on Aug. 15, 2004 Pay close attention to how you negotiate hills, ice, water and flat sufaces. You automatically adopt a different stride frequency. This enables you to move while maintaining your balance. Too long a stride on ice and you and the ice get to know each other a bit better.. See what he means now? Yes, but this would be better handled by the mech itself (you may not know the ground beneath is wet and slippery, but thet sensors in the mech's feet do) and using motion capture for something that could be done with a simple slider/wheel seems to be just too much. -- Posted by Ajay on 10:22 pm on Aug. 15, 2004 << Having no programming experiance whatsoever I personally would avoid having to get a computer tied to the control systems at all costs.  It seems like it would be nearly impossible for a mech to navigate anything but a flat surface without tripping and falling.  However, if you can get a joystick/petals/whatever to work in uneven terrain with a computer program then thats fine, it just seems harder to do.>>      This is why legged vehicles aren't as common.  Walking, especially on unstable terrain isn't a very easy task. << Personally, I think the best option (the simplist) for control would be some sort of body suit that monitored the joint angles in your body and then used a ratio to translate that into movement of the hydrolics.  The pilot would have to be suspended in midair and not touching the floor.  There would be a learning curve as the pilot would have to learn how move his limbs in such a way as to move the mech, which would have a diffrent stride, etc. as a human.>>   Implemented a while back by General Electric, this idea proved more diffiuclt than they thought and has been pretty much abandonned by walkers since.  Exoskeletons use it though.  This control system also tends to take up alot of room, as you need to have the operater practically standing, and with enough space to adequately move his limbs about.     Also, for this to effective and avoid tripping and such, the operater needs an incredibly good view of the surrounding terrain, and better feel of the ground- something that becomes very difficult when you're suspended 10 feet off the ground and the only tactile input is stiffening pistons. Not even exoskeletons go this route.  The operater walks as normal, and the suit suspends itself and walks with him. -- Posted by Earle Bishop on 11:56 pm on Aug. 15, 2004 Quote: from Archanasia on 11:13 am on Aug. 15, 2004 Personally, I think the best option (the simplist) for control would be some sort of body suit that monitored the joint angles in your body and then used a ratio to translate that into movement of the hydrolics.  The pilot would have to be suspended in midair and not touching the floor.  There would be a learning curve as the pilot would have to learn how move his limbs in such a way as to move the mech, which would have a diffrent stride, etc. as a human. MPS's system has no learning curve. The pilot never directly controls where the legs go. The machine determines expected ground pressures, and can compensate for errors after stepping down. This removes the need to move your legs in a manner free of ground interaction, and does not require you to determine how much of the time your feet need to be on the ground, or how far apart they would be. I never considered that there would be a necessity for a machine that requires you to walk everywhere you go. Compare this to a bike and a car. A bike requires your exact input and effort to move. A car requires input, but no real pressure or repetition. Mind you, I enjoyed riding my bike 10 miles to and from CSUB, but after I got my first Porsche, I never considered pedaling out of necessity. The first time you have to "walk" across any distance, you will get sick of it, too. This method is not too hard except for the hands, where you would have to deal on a much more detailed and smaller scale system of sensors.  The sensor hand thingy that I saw earlier on the site seemed to me like it was basing finger position on the overall bend of the fingers and not actual joint movements so I would be suspicious of how precise of a hand movement you could acheive with something like that.  That's just me guessing though 'cause I haven't bought one yet.  :) The P5 kinda sucks, but it does what it is supposed to: places the hand in 3D. The machine makes the decision on how exactly to get the hand to a desired spot. The operation of the fingers and positioning of the hand is up to the operator. I didn't want to suggest the idea of some nice slimfitting futurist thing that will never exist.  I was thinking of something more like the braces you might see on someone when they have a bad knee or break a bone near a joint. Check out http://www.animazoo.com The simple system you describe is clunky, and prone to wear and breakage. This is why the guys at Animazoo replaced their old system with a "futuristic" system, many years ago. -- Posted by Gideon on 3:17 pm on Aug. 21, 2004 Ah, back from my little trip. Only crashed my bike once, and it took me through a chain-link fence. Good times. Anyway, my reasoning behind things such as a fully articulated human hand, would be redundancy. Simply put, the more bells and whistles, the more rust. I don't believe that we could make an artificial arm well enough to allow for accuracy. Humans require hundreds of subconcious movements to aim a weapon. Even we don't always get it right, and it can take years to develop your skills. I also seriously doubt anyone is going to make a magazine pouch for a 30mm cannon. It would be much more simple and reliable to simply put a large ammo hopper on the weapon (no, the rounds aren't freefloating in it, its just called a hopper because it holds them) and either belt or spring feed the rounds to the chamber. In any event, weapons using self-contained rounds wouldn't be very large, because ammunition would start to eat up space, and you don't need a round the size of a baseball to kill infantry. Instead, you can use guides missiles. After all, you could make a pod of 8 missiles, and they will kill 8 tanks. Pretty much guaranteed. As for picking up other weapons, that one case is an acception. If you read the real life accounts of other soldiers throughout war, they never had the time to recover enemy weapons, especially when facing superior numbers. I urge you to read Black Hawk Down by Mark Bowden to get an idea of what the most recent urban combat is like (I know, its 10 years outdated, but until someone releases a book on Operation Enduring Freedom, there is no way for the public to know what kind of goodies and tactics are in use as of this moment). And no, don't go watch the movie. The movie was hollywood. The book is the real deal. As for construction, I doubt your machine would be cheaper than a bulldozer, and most companies don't have more than 1 of any kind of machine unless they absolutely have to. Its not very likely that they would have 2 bulldozers or 2 mechs for a 30 foot stretch of road, or to build one house. So if it breaks down, they are in the same position. Also, I have serious doubts that one could perform the work of a bulldozer. I could see them lift very heavy objects and hold them in place while workers weld or bolt it together, which would drastically decrease build time, or to load vehicles with things that other equipment cannot (such as small steel rods). And to touch on the automation part, the F-22 is said to have the type of computer I described. The vectored engines don't require direct input from the pilot, but they anticipate what they should do when he has his throttle and stick in a certain position. I don't know how it works exactly, but I doubt anyone but boeing and the pilots do. There is also the robot that honda built, which is said to anticipate based on someone's movement. if they hold out their hand and smile, for example, he assumes they want to shake his hand. I'm not saying this is feasable for you at this very moment, because both of those were multi-billion dollar projects that spanned several years, but if the legs could move correctly by simply monitoring the ground, velocity, and input from the pilot's control stick, that would cut a chunk out of the troubles of piloting it. I have no illusions about an AI system or other program that can pilot it hands free, but even commercial aircraft are controlled mostly through a computer that monitors certain elements and adjusts accordingly. -- Posted by Earle Bishop on 8:26 am on Aug. 22, 2004 Quote: from Gideon on 12:17 pm on Aug. 21, 2004 I don't believe that we could make an artificial arm well enough to allow for accuracy. I must presume then, that you are rather uninformed in the way of modern robotics. Humans require hundreds of subconcious movements to aim a weapon. Even we don't always get it right, and it can take years to develop your skills. Humans require hundreds of subconscious movements to open a door knob. A machine can very easily replicate a task (such as aiming). A machine can reach for an object (even moving) with insane speed an accuracy. I have seen basic robotics perform this feat. Not NASA, or DARPA budget robotics, but university staff. The movie was hollywood. The book is the real deal. Hollywood is crap. I like books plenty, but I prefer to ask the real McCoy. As for construction, I doubt your machine would be cheaper than a bulldozer, and most companies don't have more than 1 of any kind of machine unless they absolutely have to. That is the point. When the job-specific machine breaks, you cannot do that job until the machine is replaced, repaired, or refitted. A mech that is capable of using many tools does not have that limitation. I could see them lift very heavy objects and hold them in place while workers weld or bolt it together, which would drastically decrease build time, or to load vehicles with things that other equipment cannot (such as small steel rods). I see that as kind of limited. Why not have the machine position, hold, and weld/bolt? I don't know how it works exactly, but I doubt anyone but boeing and the pilots do. The interface is abstracted. So is ours. There is also the robot that honda built, which is said to anticipate based on someone's movement. if they hold out their hand and smile, for example, he assumes they want to shake his hand. I'm not saying this is feasable for you at this very moment, because both of those were multi-billion dollar projects that spanned several years, Yup, it took Honda like 20 years to do that. Such actions can be defined. I simply state that the complexity and time involved in such an endeavor will not have any represent able benefit for the time expended in the pursuit of said feat. but if the legs could move correctly by simply monitoring the ground, velocity, and input from the pilot's control stick, that would cut a chunk out of the troubles of piloting it. Well, not specifically the legs, but the system will use input from cameras, ground pressure, positional feedback data, pilot preference, and a record of past events that match the current situation. A mass of other stuff applies, but that is all we are giving away at present. I have no illusions about an AI system or other program that can pilot it hands free, but even commercial aircraft are controlled mostly through a computer that monitors certain elements and adjusts accordingly. A pilot is only required by FAA regulations to land one in ten flights. Having flown in a multi-million dollar 747 simulator, with FAA approved hardware ripped right out of an actual plane, there is only so much the computer can do. It can land, yes. Very well, in fact. It can very indirectly take you from point A towards point B... but all the alignment, problems, turbulence, etc, is handled by a human. The AI is not capable of working complex issues out. This is why although planes (with miles of visibility and lots of free space to correct position in) are automated, production cars (with a LOT of interaction to deal with) are still not autopilotable. -- Posted by AP on 2:27 pm on Aug. 30, 2004 Black Hawk Down is probably one of the more accureate war storys that Hollywood has made recently. the book is more accurate ofcource but Hollywood didn't bucher the story into a movie like titanic. -- Posted by ChrisDickerson on 7:31 pm on Aug. 30, 2004 Actually for acurate stories, try We Were Soldiers. The Guy who was the colonel and the reporter both refused to let Hollywood change anything about what actually happened. THey said that they would pull the plug on the entire project if Hollywood did try. -- Posted by Stormraven on 8:04 pm on Sep. 20, 2004 [I have no illusions about an AI system or other program that can pilot it hands free, but even commercial aircraft are controlled mostly through a computer that monitors certain elements and adjusts accordingly.] -------------------------------------------------------------------------------- [{A pilot is only required by FAA regulations to land one in ten flights. Having flown in a multi-million dollar 747 simulator, with FAA approved hardware ripped right out of an actual plane, there is only so much the computer can do. It can land, yes. Very well, in fact. It can very indirectly take you from point A towards point B... but all the alignment, problems, turbulence, etc, is handled by a human. The AI is not capable of working complex issues out. This is why although planes (with miles of visibility and lots of free space to correct position in) are automated, production cars (with a LOT of interaction to deal with) are still not autopilotable.}] Well...After going through 11 pages of arguments on a possible control interface(of particular interest to me since I have applied as a test pilot)I like the proposed interface so far. Going all the way back to the begining I think that joysticks in cunjunction with vocal commands are an excellent idea. "Trace-gloves" or mo-cap is also a good choice. As for the above which I selected and highlighted, How many of you are familiar with the BOLO books. They are talking about robotic ai controlled tanks with a commander on board to help render key descisions. But I go off subject. The system is going to have to be linked to some sort of real time ground mapping radar system. Especially if it going to have any possible use in a combat enviroment where the surface is undergoing constant change due to various reasons including bombardment craters etc. What sort of system due you propose to use for real time navigation? Travelling on roads etc is no problem with a GPS link and mapping, the variances in terrain in such an instance are minimal. I am also interested in the shock compensation system. -- Posted by EdZ on 3:15 am on Sep. 21, 2004 Quote: from Stormraven on 5:04 pm on Sep. 20, 2004 The system is going to have to be linked to some sort of real time ground mapping radar system. Especially if it going to have any possible use in a combat enviroment where the surface is undergoing constant change due to various reasons including bombardment craters etc. What sort of system due you propose to use for real time navigation? Some kind of LADAR system placed around the torso should be able to map terrain fairly quickly and with a high resolution. -- Posted by Merlyn on 1:15 pm on Sep. 21, 2004 This concept was briefly covered in the thread "Stumble Prevention" ie, some form of terrain mapping (LIDAR, SONAR, RADAR, etc)hooked to the navigation computer to properly adjust stride length and foot placement automatically.