Adds gym.Env to DirectMARLEnv for Gymnasium v1.0.0

source/extensions/omni.isaac.lab/config/extension.toml

@@ -1,7 +1,7 @@
 [package]
 
 # Note: Semantic Versioning is used: https://semver.org/
-version = "0.27.15"
+version = "0.27.16"
 
 # Description
 title = "Isaac Lab framework for Robot Learning"

source/extensions/omni.isaac.lab/docs/CHANGELOG.rst

@@ -1,6 +1,15 @@
 Changelog
 ---------
 
+0.27.16 (2024-11-21)
+~~~~~~~~~~~~~~~~~~~~
+
+Changed
+^^^^^^^
+
+* Changed :class:`omni.isaac.lab.envs.DirectMARLEnv` to inherit from ``Gymnasium.Env`` due to requirement from Gymnasium v1.0.0 requiring all environments to be a subclass of ``Gymnasium.Env`` when using the ``make`` interface.
+
+
 0.27.15 (2024-11-09)
 ~~~~~~~~~~~~~~~~~~~~
 

source/extensions/omni.isaac.lab/omni/isaac/lab/envs/direct_marl_env.py

@@ -34,7 +34,7 @@
 from .utils.spaces import sample_space, spec_to_gym_space
 
 
-class DirectMARLEnv:
+class DirectMARLEnv(gym.Env):
     """The superclass for the direct workflow to design multi-agent environments.
 
     This class implements the core functionality for multi-agent reinforcement learning (MARL)

source/extensions/omni.isaac.lab/setup.py

@@ -27,7 +27,7 @@
     # devices
     "hidapi",
     # reinforcement learning
-    "gymnasium==0.29.0",
+    "gymnasium",
     # procedural-generation
     "trimesh",
     "pyglet<2",

source/extensions/omni.isaac.lab_tasks/test/test_environments.py

@@ -118,7 +118,9 @@ def _check_random_actions(self, task_name: str, device: str, num_envs: int, num_
         with torch.inference_mode():
             for _ in range(num_steps):
                 # sample actions according to the defined space
-                actions = sample_space(env.single_action_space, device=env.unwrapped.device, batch_size=num_envs)
+                actions = sample_space(
+                    env.unwrapped.single_action_space, device=env.unwrapped.device, batch_size=num_envs
+                )
                 # apply actions
                 transition = env.step(actions)
                 # check signals

source/extensions/omni.isaac.lab_tasks/test/test_multi_agent_environments.py

@@ -105,7 +105,7 @@ def _check_random_actions(self, task_name: str, device: str, num_envs: int, num_
 
         # this flag is necessary to prevent a bug where the simulation gets stuck randomly when running the
         # test on many environments.
-        env.sim.set_setting("/physics/cooking/ujitsoCollisionCooking", False)
+        env.unwrapped.sim.set_setting("/physics/cooking/ujitsoCollisionCooking", False)
 
         # reset environment
         obs, _ = env.reset()
@@ -116,7 +116,9 @@ def _check_random_actions(self, task_name: str, device: str, num_envs: int, num_
             for _ in range(num_steps):
                 # sample actions according to the defined space
                 actions = {
-                    agent: sample_space(env.action_spaces[agent], device=env.unwrapped.device, batch_size=num_envs)
+                    agent: sample_space(
+                        env.unwrapped.action_spaces[agent], device=env.unwrapped.device, batch_size=num_envs
+                    )
                     for agent in env.unwrapped.possible_agents
                 }
                 # apply actions