Open luismuschal opened 6 months ago
This isn't a mujoco-py issue, you would be better off asking on https://github.com/google-deepmind/mujoco or (more likely) https://github.com/ARISE-Initiative/robosuite.
Also you are using a version of MuJoCo that's several years old by this point.
Environment:
Description: I am observing different values in the diagonal elements of the mass matrices computed by MuJoCo and Pinocchio when using robosuite. This offset is always constant. This occurs with both a Schunk Robot and a 1-DOF rotational joint setup. However, the torques due to nonlinear effects (
sim.data.qfrc_bias
) are consistent between the systems. When I use the following xml-file created from robosuite:and the .urdf for pinocchio:
I get a value of [[5.02]] using the following code from robosuite (base_controller.py)
and [[0.02]] for the pinocchio implementation:
Expected vs. Actual Results: