tud-airlab / mppi-isaac

Model Predictive Path Integral Control using isaacgym for rollouts, gpu-accelerated
166 stars 19 forks source link

Ft multi agent support #34

Closed c-salmi closed 1 year ago

c-salmi commented 1 year ago

To support the centralized control of multiple robots through a single mppi, it was necessary to change quite a few thing in the structure of the IsaacGymWrapper.

cpezzato commented 1 year ago

Nice!

I have checked most of the examples but it looks like the boxer in boxer.py does not move in pybullet due to friction of the caster wheels. The samples are fine though in isaacgym and they can move

c-salmi commented 1 year ago

There was a bug in the wrapper that caused the robot positions in the cost function for the boxer example to be zero. I don't think the friction of the caster joint was the issue, but it should be fixed now.