Bioloid robot self-balances at The University of Plymouth
Bioloid robot self-balances at The University of Plymouth
by Lexa
Sat Oct 18, 2008 2:01 am
Hello,
I have done my msc project in robotics on a bioloid robot. The point was to develop a motion controller for it.
I am now publishing my works and outcomes. You can find my thesis, sources codes and other things related here:
http://plymouth.tinigrifi.eu/prjThesis.php
My project comprised:
- A physical simulator half done. The simulator is made in java and use the PhysX engine. Physics measurement have been taken in the work of the IDMEC in Portugal.
- A bipedal motion controller involving a reflexive controller (inspiration from RunBot project), a leg abstraction (inspiration from Nimro) and a primitive controller.
One of the main outcomes is that the use of compliance settings in AX12 actuators have shown ability to produce self-generated oscillations in the lateral plan. Grossly, the elasticity/damping/spring effect coming from the torque compliance tweaking let the robot dynamics act on hip and ankle roll joints (and also on all other, but in a lesser way).
The walk is not always stable but is quite fast and does not rely on static postures (no forced bending on one side to another). Knees are of course still bended due to the Bioloid Humanoid structure (knee slightly backward). I also pointing out the fact that parameters have not been explored AT ALL. I just tried some values and it worked. So there is of course a lot of possible improvement.
Everything is detailed in resources available on my website.
Plus, I realised a short video on it:
I also would like to thank the community here and robosavvy as I used some resources available on this site and forum during my researches
.
I have done my msc project in robotics on a bioloid robot. The point was to develop a motion controller for it.
I am now publishing my works and outcomes. You can find my thesis, sources codes and other things related here:
http://plymouth.tinigrifi.eu/prjThesis.php
My project comprised:
- A physical simulator half done. The simulator is made in java and use the PhysX engine. Physics measurement have been taken in the work of the IDMEC in Portugal.
- A bipedal motion controller involving a reflexive controller (inspiration from RunBot project), a leg abstraction (inspiration from Nimro) and a primitive controller.
One of the main outcomes is that the use of compliance settings in AX12 actuators have shown ability to produce self-generated oscillations in the lateral plan. Grossly, the elasticity/damping/spring effect coming from the torque compliance tweaking let the robot dynamics act on hip and ankle roll joints (and also on all other, but in a lesser way).
The walk is not always stable but is quite fast and does not rely on static postures (no forced bending on one side to another). Knees are of course still bended due to the Bioloid Humanoid structure (knee slightly backward). I also pointing out the fact that parameters have not been explored AT ALL. I just tried some values and it worked. So there is of course a lot of possible improvement.
Everything is detailed in resources available on my website.
Plus, I realised a short video on it:
phpBB [media]
I also would like to thank the community here and robosavvy as I used some resources available on this site and forum during my researches
.