habitat.gym.gym_wrapper module

Classes

class HabGymWrapper

Functions

def _is_continuous(original_space: gym.spaces.space.Space) -> bool
returns true if the original space is only suitable for continuous control
def _recursive_continuous_size_getter(original_space: gym.spaces.space.Space, low: typing.List[float], high: typing.List[float])
Returns the size of a continuous action vector from a habitat environment action space
def continuous_vector_action_to_hab_dict(original_action_space: gym.spaces.space.Space, vector_action_space: gym.spaces.box.Box, action: numpy.ndarray) -> typing.Dict[str, typing.Any]
Converts a np.ndarray vector action into a habitat-lab compatible action dictionary.
def create_action_space(original_space: gym.spaces.space.Space) -> gym.spaces.space.Space
Converts a habitat task action space into a either continuous (Box) or discrete (Discrete) gym.space.
def filter_observation_space(obs_space, limit_keys)
def smash_observation_space(obs_space, limit_keys)

Data

KEYFRAME_OBSERVATION_KEY = 'keyframe'
TYPE_CHECKING = False
pygame = None