Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Feature] Support wrapping IsaacLab environments with GymEnv #2380

Merged
merged 4 commits into from
Aug 9, 2024
Merged

[Feature] Support wrapping IsaacLab environments with GymEnv #2380

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
ffbcda4
add option for converting actions to numpy
yu-fz Aug 7, 2024
132072a
apply formatting
yu-fz Aug 7, 2024
1fb77e6
fix kwarg passing
yu-fz Aug 7, 2024
73596a8
move kwarg pop to env base
yu-fz Aug 8, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions torchrl/envs/common.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3039,6 +3039,7 @@ def __init__(

self._constructor_kwargs = kwargs
self._check_kwargs(kwargs)
self._convert_actions_to_numpy = kwargs.pop("convert_actions_to_numpy", True)
self._env = self._build_env(**kwargs) # writes the self._env attribute
self._make_specs(self._env) # writes the self._env attribute
self.is_closed = False
Expand Down
6 changes: 4 additions & 2 deletions torchrl/envs/gym_like.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -172,6 +172,7 @@ class GymLikeEnv(_EnvWrapper):
def __new__(cls, *args, **kwargs):
self = super().__new__(cls, *args, _batch_locked=True, **kwargs)
self._info_dict_reader = []

return self

def read_action(self, action):
Expand Down Expand Up @@ -289,7 +290,8 @@ def read_obs(

def _step(self, tensordict: TensorDictBase) -> TensorDictBase:
action = tensordict.get(self.action_key)
action_np = self.read_action(action)
if self._convert_actions_to_numpy:
action = self.read_action(action)

reward = 0
for _ in range(self.wrapper_frame_skip):
Expand All @@ -300,7 +302,7 @@ def _step(self, tensordict: TensorDictBase) -> TensorDictBase:
truncated,
done,
info_dict,
) = self._output_transform(self._env.step(action_np))
) = self._output_transform(self._env.step(action))

if _reward is not None:
reward = reward + _reward
Expand Down
5 changes: 5 additions & 0 deletions torchrl/envs/libs/gym.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -645,6 +645,11 @@ class GymWrapper(GymLikeEnv, metaclass=_AsyncMeta):
allow_done_after_reset (bool, optional): if ``True``, it is tolerated
for envs to be ``done`` just after :meth:`~.reset` is called.
Defaults to ``False``.
convert_actions_to_numpy (bool, optional): if ``True``, actions will be
converted from tensors to numpy arrays and moved to CPU before being passed to the
env step function. Set this to ``False`` if the environment is evaluated
on GPU, such as IsaacLab.
Defaults to ``True``.

Attributes:
available_envs (List[str]): a list of environments to build.
Expand Down
Loading