Open aleada23 opened 4 months ago
Yuval's suggestion under your previous post would probably work for you. If you have your own way of updating or setting qpos/qvel, you can just call individual sub-steps from mj_step as needed.
I recommend reading up of the simulation loop section in the docs, especially the breakdown of mj_step1
. You can check the source code of the substeps inside mj_step1
(e.g. for mj_fwdPosition) or their summaries in the API documentation, if you need to go even more fine grained.
Thank you, I'll check the documentation that you provided.
Discussed in https://github.com/google-deepmind/mujoco/discussions/1533