garage.envs.dm_control

Wrappers for the DeepMind Control Suite.

See https://github.com/deepmind/dm_control

class DMControlEnv(env, name=None)

Bases: garage.Environment

Binding for dm_control <https://arxiv.org/pdf/1801.00690.pdf>.

action_space

The action space specification.

Type:akro.Space

observation_space

The observation space specification.

Type:akro.Space

spec

The environment specification.

Type:EnvSpec

render_modes

A list of string representing the supported render modes.

Type:list

classmethod from_suite(cls, domain_name, task_name)

Create a DmControl task given the domain name and task name.

Parameters:

• domain_name (str) – Domain name
• task_name (str) – Task name

Returns:

the dm_control task environment

Return type:

dm_control.suite.Task

reset(self)

Resets the environment.

Returns:

The first observation conforming to

observation_space.

dict: The episode-level information.

Note that this is not part of env_info provided in step(). It contains information of he entire episode， which could be needed to determine the first action (e.g. in the case of goal-conditioned or MTRL.)

Return type:numpy.ndarray

step(self, action)

Steps the environment with the action and returns a EnvStep.

Parameters:action (object) – input action Returns:The environment step resulting from the action. Return type:EnvStep Raises:RuntimeError – if step() is called after the environment has been constructed and reset() has not been called.

render(self, mode)

Render the environment.

Parameters:mode (str) – render mode. Returns:if mode is ‘rgb_array’, else return None. Return type:np.ndarray Raises:ValueError – if mode is not supported.

visualize(self)

Creates a visualization of the environment.

close(self)

Close the environment.

class DmControlViewer

Bases: dm_control.viewer.application.Application

render(self)

launch(self, environment_loader, policy=None)

close(self)