My two proposed changes are not related. Feel if you want me to submit them separately.
The first change in pybulletgym/envs/mujoco/envs/env_bases.py was already submitted for robotschool environments. This just reflects the same change due to a renaming for the mujoco envs.
The second change is due to the fact that the observations returned should be a 1-D array, not a 2x2 array. The current implementation fails if you combine the environment with existing agents from stable_baselines. Also, this change makes things more consistent.
My two proposed changes are not related. Feel if you want me to submit them separately.
The first change in
pybulletgym/envs/mujoco/envs/env_bases.py
was already submitted for robotschool environments. This just reflects the same change due to a renaming for the mujoco envs.The second change is due to the fact that the observations returned should be a 1-D array, not a 2x2 array. The current implementation fails if you combine the environment with existing agents from stable_baselines. Also, this change makes things more consistent.