observation_space return None

[izvvv]

state_dim = env.observation_space.shape[0]
action_dim = env.action_space.shape[0]
max_action = float(env.action_space.high[0])

parking-v0
env.observation_space.shape return none lead to error
how to deal with it?

[eleurent]

This is because for ParkingEnv is compatible with training with HER (Hindsight Experience Replay), which requires it to be (see first warning) a gym.GoalEnv, and GoalEnv environments use Dict as observation spaces, not Boxes/arrays, so env.observation.space.shape wont work, you first need to get its fields, and then get their shape.

For example,

print(env.observation_space['observation'].shape)
print(env.observation_space['achieved_goal'].shape)
print(env.observation_space['desired_goal'].shape)