Physics Walkers

OK this idea has been rolling around in the back of my head for a while. I did some tests today and I believe it is possible. Using a linked set of prims that shift COG by moving some of the child prims, can a physics walker be made? I was thinking that the 'feet' can sense collision and adjust their rotation to flatten to ground contour. Once a foot hits ground and adjusts, a link message can activate the COG shift and turn the object physics again. When the next collision occurs, the set adjust again and so on. The set would basically follow the cycle: fall, sense collision, go non-physics (to do setpos and setrot of child prims), adjust COG weight, go physics, repeat. Thoughts?

I'm certain this can be done. I was thinking of doing something like this by using an evolutionary algorithm to "evolve" the behavior for each leg to move the construct. It's been done before, so was interested in trying to repeat the experiment here.

Hmm...

Well, I'd say its POSSIBLE.. but it would take alot of drudge work =) Also, many many scripts. (Cant forget the 16kb limit). If you can get it working, though, great! More power to ya.

Good luck.

Here is something to inspire you while you're at it. Check the Natural Motion site for more details on this clip.

http://www.naturalmotion.com/files/simplewalkevolution.mpg

Wouldn't it be easier to start with something with 4 legs, so it doesn't fall over as easily on the uneven terrain of SL?

Little

Now see. Little Tigereye offered up a great clip! You guys have to see that. I was just suggesting the idea, not the whole model. Sure 4 legs would probably be a good starting point.

While this is slightly off topic, I've wanted a way to install Linden-made animations on user created bipedal objects for a long time. Either that or give us the ability to attach things up to 20 meters away from our AVs.

A walking mecha will someday be mine, I say.

I sure would like to see a walking biped in Secondlife
But learning it to walk with an evolutionary algoritm seems a little too ambitious for LSL.

Would LSL provide enough reaction time? Would Havok provide enough precision?

My guess is no and no and I really doubt it will ever be done
But might be worth a shot. Maybe you won't get it too work, but you'll learn enough new things and you might get another cool idea from the stuff you learned