| View previous topic :: View next topic |
| Author |
Message |
robotjay co-admin
Joined: 01 Aug 2006 Posts: 225 Location: Nebraska, USA
|
Posted: Mon Aug 06, 2007 6:28 pm Post subject: Simple (Not so simple?) Biped |
|
|
This is a link to a YouTube video of a simple looking, but very complex biped called Runbot. I think this link was sent to me by someone who knows someone who had it sent to them by Barry Carter, but either way, here it is:
http://www.youtube.com/watch?v=hYTnBd063MQ
Dynamic gait learning. Beautiful.
-Jay _________________ "Nothing is fool-proof; For we fools are ingenious and will find a way." |
|
| Back to top |
|
 |
kbb
Joined: 01 Jun 2007 Posts: 180
|
|
| Back to top |
|
 |
jamma
Joined: 30 Mar 2006 Posts: 24
|
Posted: Tue Aug 14, 2007 12:52 am Post subject: |
|
|
RunBot looks very cool, and capitalizes on a passive leg swing for part of the gait. I've always thought this is the way bipeds should go (vs the "always on" approach of current bipeds). That, of course, and a learning mechanism which RunBot also has. Very informative write-up here:
http://compbiol.plosjournals.org/perlserv/?request=get-document&doi=10.1371/journal.pcbi.0030134
Sadly, the uphill gait change involves a bit of a cheat: they use an IR sensor to distinguish the normally black flat footpath from the white inclined ramp. I'd think with the proper IMU integration, this could be "sensed" rather than "seen" (and allow for the compensation of more complex terrain, too.) I believe Mike was working on a Kalman filter for IMU sensor integration. That seems the obvious next step for a robot like this.
Overall, this work is an excellent springboard from which the biped enthusiast can work. Any takers? Sadly, my programming skills are not nearly up to snuff (although the mechanics are well within my capabilities.) |
|
| Back to top |
|
 |
|