Sumotori

[MYKL]

There's no lack of academics exploring biped locomotion. The fascinating thing about Sumotori is that the it was done by a hacker with some unbelievable constraints. Just the type to be able to get a real robot to do the same in a limited timeframe.

... take our humanoids to a higher level of stability and movement.
Even with advanced control our real humanoid robots would be limited to tai-chi and yoga.

I'm extremely impressed with both.

What I'd really like to see is a sim program that allows you to bring your own mechanical construct into it and program the movements based on servo control only (with the same RL constrains as purchaseable servos). So I could design whatever machine I wanted in Solidworks , Inventor or whatever and then move the construct over and use the simulators servos to move/program the machine. The simulator would supply the physics as in a real word situation. Perhaps you could even have varied surfaces to test your new machine on. I see that the controls and dynamics (the brains) for movement are far ahead of the mechanics. There are guys employing some awesome IK stuff in they're bots. I have been buliding animatronic style puppeted stuff for some time and I believe I could design an impressive machine. I'd just like to have some help with the control. I am a bit undereducated as far as electronics are concerned so a user friendly GUI for designing and programming ANY constrcut would be a godsend.

[limor]

Just noticed that Sumotori is still up and about and that ZMP self stabilizing humanoid done by Peter Dutch over a weekend hackathon in some 7 years ago, is still in some ways state of the art in robotics simulation.

https://www.youtube.com/watch?v=40z5BTh ... A4E82C472A

phpBB [media]

https://play.google.com/store/apps/deta ... tori&hl=en