Update unity_pettingzoo_base_env.py

[AmineAndam04]

Proposed change(s)

in the the _update_action_spaces method of the UnityPettingzooBaseEnv class, you force the actions to be of int32 and in the [-1,1] range (see the code below). This constraint doesn't apply to all scenarios

act_spec.continuous_size > 0: c_space = spaces.Box( -1, 1, (act_spec.continuous_size,), dtype=np.int32 )

It makes more sense to use the same logic as in _update_observation_spaces.

obs_spaces = tuple( spaces.Box( low=-np.float32(np.inf), high=np.float32(np.inf), shape=spec.shape, dtype=np.float32, )

We can generalize to all scenarios by using:

if act_spec.continuous_size > 0: c_space = spaces.Box( -np.float32(np.inf), np.float32(np.inf), (act_spec.continuous_size,), dtype=np.float32 ) 

Types of change(s)

• [x ] Bug fix
• New feature
• Code refactor
• Breaking change
• Documentation update
• Other (please describe)

Checklist

• Added tests that prove my fix is effective or that my feature works
• Updated the changelog (if applicable)
• Updated the documentation (if applicable)
• Updated the migration guide (if applicable)

Other comments

[CLAassistant]

All committers have signed the CLA.

[alexander-zap]

+1, ran into the same problem. This is especially problematic when only using the mlagents-envs package for environment interfacing, but using own training algorithms (e.g., StableBaselines3).