Acquire_actor_root_state_tensor documentation shows incorrect dimensions

The documentation for acquire_actor_root_state_tensor states:

Retrieves buffer for Actor root states. Buffer has shape (num_environments, num_actors * 13). State for each actor…

However, my testing shows that the correct shape is (num_environments * num_actors, 13). Simply create a simulation with n envs and m actors to verify.

2 Likes