5

I'm working on a legged robot and generating joint torques. Basically the robot seems to be statically stable to some extend. The robot goes instable if the center of pressure moves to the border of the feet. I'm looking for some method to move away the center of pressure from the feet edges after having calculated my joint torques. In Sentis thesis ( http://ai.stanford.edu/~lsentis/files/Thesis-Sentis-2007.pdf ) , it is mentioned that he somehow manages to cancel out the internal forces to keep the feet flat against the supporting surfaces.

Does anyone has got experience in dealing with internal forces? As far as I understood the literature one can modify the nullspace of the calculated torques to achieve that the COP remains in the geometrical center of the considered foot. I'm looking for methods apart from the virtual linkage model as it did not seem to work for me or someone with whom I could discuss the virtual linkage model described in ( http://ai.stanford.edu/~lsentis/files/tro-2010.pdf ) as I might not have it understood it correctly.

Shahbaz
  • 3,250
  • 1
  • 20
  • 38
  • I have only just started looking into operational space control and in turn only just started reading through Sentis' thesis. I was looking forward to an answer to this question! He does mention in section 2.3 that they are working on controllers for internal forces (at the time of writing). You might like to try and email him directly to find out about their progress. I have had some great replies from emailing the people directly about their work. Good luck! – law Nov 16 '13 at 23:59

1 Answers1

1

Sentis' lab at UT Austin has open source code that implements many of the control topics mentioned in his thesis. You can check it out here: https://github.com/liangfok/controlit

Specifically for the internal forces monitoring, you can checkout: https://github.com/liangfok/controlit/blob/b8f3ba056b5329e26b9519c8923c43ae58f0dcd9/controlit_core/include/controlit/VirtualLinkageModel.hpp

Barrett Ames
  • 31
  • 4