PettingZoo Environments

While PettingZoo may not provide the fastest API for multi agent reinforcement learning environments, it is the most widely supported and well maintained API. For that reason, PyFlyt provides various default PettingZoo environments for testing reinforcement learning algorithms. All environments are by default ParallelEnvs. Usage is no different to how PettingZoo environments are initialized:

from PyFlyt.pz_envs import MAQuadXHoverEnv
env = MAQuadXHoverEnv(render_mode="human")
observations, infos = env.reset(seed=42)

while env.agents:
    # this is where you would insert your policy
    actions = {agent: env.action_space(agent).sample() for agent in env.agents}

    observations, rewards, terminations, truncations, infos = env.step(actions)
env.close()