OmniGibson: a platform for accelerating Embodied AI research built upon NVIDIA's Omniverse engine. Join our Discord for support: https://discord.gg/bccR5vGFEx
Recently we had to implement a compute_no_op_action function on controllers in order to be able to make robots stay still. There also is a compute_no_op_goal function which is used to populate the initial goal after a controller reset. But the latter does not always do what you expect (e.g. see multifinger) and might even be redundant since the goal could simply be computed using the no_op action.
cgokmen
commented
3 days ago
Recently we had to implement a compute_no_op_action function on controllers in order to be able to make robots stay still. There also is a compute_no_op_goal function which is used to populate the initial goal after a controller reset. But the latter does not always do what you expect (e.g. see multifinger) and might even be redundant since the goal could simply be computed using the no_op action.