I have been haunted with this simple thought. I am finally just now able to get my thoughts defined… Not sure if it’s possible though.
Ok so we have this precision software that handles all of the IK for a 6 leg robot based on some remote control method of input. The walking gaits, body moves, etc, all are working flawlessly, so what’s next?
What if we tied one of the body move variables to the same joystick as the motion. The result could be the bot noses up when backing, noses down when attacking, dip and sway into turns. Also by implementing a ghost from the joystick positions in software the dip could dissipate after time. What I mean by ghost is, make variable B (body move) follow variable A (walk direction). When the A value changes abruptly it goes through quickly to B, but if the A value stays the same, B slowly goes back to where it was. Mostly for Xan and Zenta, have you guys thought of something like this? If no one else chimes in, I’m taking credit for thinking it up. lol 8)
I like the idea as it would make the movements appear more fluid like. After a forward walk sequence has been initiated, having the body axis slowly return back to a level plane from a nose down position would be a cool effect. A delay of some sort could be used to set the speed at which each axis recovers.
Ding dong…
I’ve thought alot about this, and I believe others have that too. Especially for my A-Pod I’m planning to do something similar, for making the walking part more “realistic”. I was planning to post more info about my progress when I’m finished with the hardware modifications on the DIY XBee remote.
Hi,
Wouldn’t it be more interesting to incorporate tilt/accelerometer sensors on the bot to give a more artificial effect.
I’m using a tilt/accelerometer now to balance my hex, it holds the body level at all times side to side and front to back while walking side hill or up or down hill or angles and under PS2 control.
I’ve observed some interesting behavior like when the bot is at a stand still it’s not frozen in place but is continuously moving ever so slightly and may stop now and then, this is giving the bot the appearance of actually balancing itself by changing to it’s environment.
As for leaning into a turn or into forward direction it’s just a matter of reversing the code. I’m using only one sensor right now for two directions and can see the potential for more sensors of this type that will give the bot an alive appearance but with full PS2 or whatever controller input.
It would be like taking a live beetle and installing a remote control in it.
I’ve done self leveling decks years ago, so that’s not as impressive to me. But to have a hexapod show expression when walking with body moves tied into translation and rotation wouldbe jaw dropping cool
I was just thinking that if by doing the opposite of self leveling that say when you go forward an accelerometer can measure that movment and adjust accordinly you would get the effect that you want plus a more natural rather than a set " lean/tilt " as per joystick position.
Kind of like if you walk forward gradually from slow to fast you wouldn’t necessarily need the bot to lean into it but if you slammed the stick forward the bot would lean in hard and then slowly back off as speed increased.
Or say if you were going down hill your bot wouldn’t be leaning back but if it slipped it would start to lean back.
I’m already using some code mods in my bot now but it is easier to talk what I want to do than it is to mod the code but I’m working on it.
I know some of you guys spit this stuff out in your sleep but I quit doing that…I was loosing weight, starving,dehydrated and sleepless…plus for some reason my wife started asking me who I was at night time…yuk…yuk.
Sorry to disappoint you but I’ve been thinking about this for a while. I even can remember discusting this with Zenta.
I defiantly agree that combining multiple inputs like rotating/shifting the body and walking will give it a more charisma. But to get the total WOW factor I was thinking of a bit more complex methods then only combining inputs.
I think that with 2.0 we’ve got a good engine for the controls. To get more “life†we should build a AI layer on top of the engine. Think about vision or voice response. They could be separate modules that can be added or improved bit by bit. The AI layer will also take care of the emotions shown by the walker. This is one of the things I really want to do in the future!
Work and private stuff are currently holding me from finishing this part. But for what it counts, I’m working on this…
