V0.7.0: Gymnasium Upgrade. Multiple teleOP controllers. MuJoCo upgrade

README.md

@@ -7,15 +7,14 @@ License :: Under Apache License, Version 2.0 (the "License"); you may not use th
 
 <!-- # RoboHive -->
 
-![PyPI](https://img.shields.io/pypi/v/robohive)
+[**[Homepage]**](https://sites.google.com/view/robohive) &ensp; [**[White Paper]**](https://arxiv.org/abs/2310.06828) &ensp; [![Documentation](https://img.shields.io/static/v1?label=Wiki&message=Documentation&color=<green)](https://github.com/vikashplus/robohive/wiki) ![PyPI](https://img.shields.io/pypi/v/robohive)
 ![PyPI - License](https://img.shields.io/pypi/l/robohive)
 [![Downloads](https://pepy.tech/badge/robohive)](https://pepy.tech/project/robohive)
 [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1rdSgnsfUaE-eFLjAkFHeqfUWzAK8ruTs?usp=sharing)
 [![Slack](https://img.shields.io/badge/Slack-4A154B?style=for-the-badge&logo=slack&logoColor=white)](https://robohiveworkspace.slack.com)
-[![Documentation](https://img.shields.io/static/v1?label=Wiki&message=Documentation&color=<green)](https://github.com/vikashplus/robohive/wiki)
 
 ![RoboHive Social Preview](https://github.com/vikashplus/robohive/assets/12837145/04aff6da-f9fa-4f5f-abc6-cfcd70c6cd90)
-`RoboHive` is a collection of environments/tasks simulated with the [MuJoCo](http://www.mujoco.org/) physics engine exposed using the OpenAI-Gym API.
+`RoboHive` is a collection of environments/tasks simulated with the [MuJoCo](http://www.mujoco.org/) physics engine exposed using the OpenAI-Gym API. Its compatible with any gym-compatible agents training framework ([Stable Baselines](https://stable-baselines3.readthedocs.io/en/master), [RLlib](https://docs.ray.io/en/latest/rllib/index.html), [TorchRL](https://pytorch.org/rl/), [AgentHive](https://sites.google.com/view/robohive/baseline), etc)
 
 # Getting Started
    Getting started with RoboHive is as simple as -

robohive/__init__.py

@@ -7,7 +7,7 @@
 
 
 # RoboHive version
-__version__ = "0.6.0"
+__version__ = "0.7.0"
 
 
 # Check for RoboHive initialization
@@ -16,63 +16,78 @@
 
 
 # Register RoboHive Envs
-import gym
-_current_gym_envs = gym.envs.registration.registry.env_specs.keys()
+from robohive.utils import gym
+from robohive.utils.implement_for import implement_for
+
+#TODO: check versions
+@implement_for("gym", None, "0.24")
+def gym_registry_specs():
+    return gym.envs.registry.env_specs
+
+@implement_for("gym", "0.24", None)
+def gym_registry_specs():
+    return gym.envs.registry
+
+@implement_for("gymnasium")
+def gym_registry_specs():
+    return gym.envs.registry
+
+_current_gym_envs = gym_registry_specs().keys()
 _current_gym_envs = set(_current_gym_envs)
 robohive_env_suite = set()
 
 # Register Arms Suite
 import robohive.envs.arms # noqa
-robohive_arm_suite = set(gym.envs.registration.registry.env_specs.keys())-robohive_env_suite-_current_gym_envs
+robohive_arm_suite = set(gym_registry_specs().keys())-robohive_env_suite-_current_gym_envs
 robohive_arm_suite = set(sorted(robohive_arm_suite))
 robohive_env_suite = robohive_env_suite | robohive_arm_suite
 
 # Register MyoBase Suite
 import robohive.envs.myo.myobase # noqa
-robohive_myobase_suite = set(gym.envs.registration.registry.env_specs.keys())-robohive_env_suite-_current_gym_envs
+robohive_myobase_suite = set(gym_registry_specs().keys())-robohive_env_suite-_current_gym_envs
 robohive_env_suite  = robohive_env_suite | robohive_myobase_suite
 robohive_myobase_suite = sorted(robohive_myobase_suite)
 
 # Register MyoChal Suite
 import robohive.envs.myo.myochallenge # noqa
-robohive_myochal_suite = set(gym.envs.registration.registry.env_specs.keys())-robohive_env_suite-_current_gym_envs
+robohive_myochal_suite = set(gym_registry_specs().keys())-robohive_env_suite-_current_gym_envs
 robohive_env_suite  = robohive_env_suite | robohive_myochal_suite
 robohive_myochal_suite = sorted(robohive_myochal_suite)
 
 # Register MyoDM Suite
 import robohive.envs.myo.myodm # noqa
-robohive_myodm_suite = set(gym.envs.registration.registry.env_specs.keys())-robohive_env_suite-_current_gym_envs
+robohive_myodm_suite = set(gym_registry_specs().keys())-robohive_env_suite-_current_gym_envs
 robohive_env_suite  = robohive_env_suite | robohive_myodm_suite
 robohive_myodm_suite = sorted(robohive_myodm_suite)
 
 # Register FM suite
 import robohive.envs.fm # noqa
-robohive_fm_suite = set(gym.envs.registration.registry.env_specs.keys())-robohive_env_suite-_current_gym_envs
+robohive_fm_suite = set(gym_registry_specs().keys())-robohive_env_suite-_current_gym_envs
 robohive_env_suite  = robohive_env_suite | robohive_fm_suite
 robohive_fm_suite = sorted(robohive_fm_suite)
 
 # Register Hands Suite
 import robohive.envs.hands # noqa
 # import robohive.envs.tcdm # noqa # WIP
-robohive_hand_suite = set(gym.envs.registration.registry.env_specs.keys())-robohive_env_suite-_current_gym_envs
+robohive_hand_suite = set(gym_registry_specs().keys())-robohive_env_suite-_current_gym_envs
 robohive_env_suite  = robohive_env_suite | robohive_hand_suite
 robohive_hand_suite = sorted(robohive_hand_suite)
 
 # Register Claw suite
 import robohive.envs.claws # noqa
-robohive_claw_suite = set(gym.envs.registration.registry.env_specs.keys())-robohive_env_suite-_current_gym_envs
+robohive_claw_suite = set(gym_registry_specs().keys())-robohive_env_suite-_current_gym_envs
 robohive_env_suite  = robohive_env_suite | robohive_claw_suite
 robohive_claw_suite = sorted(robohive_claw_suite)
 
 # Register Multi-task Suite
 import robohive.envs.multi_task # noqa
-robohive_multitask_suite = set(gym.envs.registration.registry.env_specs.keys())-robohive_env_suite-_current_gym_envs
+robohive_multitask_suite = set(gym_registry_specs().keys())-robohive_env_suite-_current_gym_envs
 robohive_env_suite  = robohive_env_suite | robohive_multitask_suite
 robohive_multitask_suite = sorted(robohive_multitask_suite)
 
 # Register Locomotion Suite
 import robohive.envs.quadrupeds # noqa
-robohive_quad_suite = set(gym.envs.registration.registry.env_specs.keys())-robohive_env_suite-_current_gym_envs
+robohive_quad_suite = set(gym_registry_specs().keys())-robohive_env_suite-_current_gym_envs
 robohive_env_suite  = robohive_env_suite | robohive_quad_suite
 robohive_quad_suite = sorted(robohive_quad_suite)
 

robohive/envs/arms/__init__.py

@@ -5,7 +5,9 @@
 License :: Under Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
 ================================================= """
 
-from gym.envs.registration import register
+# from gym.envs.registration import register
+from robohive.utils import gym; register=gym.register
+
 from robohive.envs.env_variants import register_env_variant
 import os
 curr_dir = os.path.dirname(os.path.abspath(__file__))

robohive/envs/arms/fetch/assets/fetch_reach_v0.xml

@@ -40,4 +40,8 @@
 
     </worldbody>
 
+    <keyframe>
+        <key name="pick_pose" qpos='-0.000445298 0 0.4334 0.0803 -0.906015 -0.0314 0.87789 0 1.62 0 0.0315 0.0315'/>
+    </keyframe>
+
 </mujoco>

robohive/envs/arms/pick_place_v0.py

@@ -6,7 +6,7 @@
 ================================================= """
 
 import collections
-import gym
+from robohive.utils import gym
 import numpy as np
 
 from robohive.envs import env_base
@@ -70,6 +70,12 @@ def _setup(self,
         self.randomize = randomize
         self.geom_sizes = geom_sizes
 
+        # Save body init pos
+        self.init_body_pos = {}
+        for body in ["obj0", "obj1", "obj2"]:
+            bid = self.sim.model.body_name2id(body)
+            self.init_body_pos[body] = self.sim.model.body_pos[bid].copy()
+
         super()._setup(obs_keys=obs_keys,
                        weighted_reward_keys=weighted_reward_keys,
                        reward_mode=reward_mode,
@@ -109,7 +115,7 @@ def get_reward_dict(self, obs_dict):
         rwd_dict['dense'] = np.sum([wt*rwd_dict[key] for key, wt in self.rwd_keys_wt.items()], axis=0)
         return rwd_dict
 
-    def reset(self):
+    def reset(self, **kwargs):
 
         if self.randomize:
             # target location
@@ -119,19 +125,19 @@ def reset(self):
             # object shapes and locations
             for body in ["obj0", "obj1", "obj2"]:
                 bid = self.sim.model.body_name2id(body)
-                self.sim.model.body_pos[bid] += self.np_random.uniform(low=[-.010, -.010, -.010], high=[-.010, -.010, -.010])# random pos
+                self.sim.model.body_pos[bid] = self.init_body_pos[body] + self.np_random.uniform(low=[-.010, -.010, -.010], high=[-.010, -.010, -.010])# random pos
                 self.sim.model.body_quat[bid] = euler2quat(self.np_random.uniform(low=(-np.pi/2, -np.pi/2, -np.pi/2), high=(np.pi/2, np.pi/2, np.pi/2)) ) # random quat
 
                 for gid in range(self.sim.model.body_geomnum[bid]):
                     gid+=self.sim.model.body_geomadr[bid]
-                    self.sim.model.geom_type[gid]=self.np_random.randint(low=2, high=7) # random shape
+                    self.sim.model.geom_type[gid]=self.np_random.choice([2,3,4,5,6,7]) # random shape
                     self.sim.model.geom_size[gid]=self.np_random.uniform(low=self.geom_sizes['low'], high=self.geom_sizes['high']) # random size
                     self.sim.model.geom_pos[gid]=self.np_random.uniform(low=-1*self.sim.model.geom_size[gid], high=self.sim.model.geom_size[gid]) # random pos
                     self.sim.model.geom_quat[gid]=euler2quat(self.np_random.uniform(low=(-np.pi/2, -np.pi/2, -np.pi/2), high=(np.pi/2, np.pi/2, np.pi/2)) ) # random quat
                     self.sim.model.geom_rgba[gid]=self.np_random.uniform(low=[.2, .2, .2, 1], high=[.9, .9, .9, 1]) # random color
             self.sim.forward()
 
-        obs = super().reset(self.init_qpos, self.init_qvel)
+        obs = super().reset(self.init_qpos, self.init_qvel, **kwargs)
         return obs
 
     # def viewer_setup(self):

robohive/envs/arms/push_base_v0.py

@@ -6,7 +6,7 @@
 ================================================= """
 
 import collections
-import gym
+from robohive.utils import gym
 import numpy as np
 
 from robohive.envs import env_base
@@ -103,8 +103,8 @@ def get_reward_dict(self, obs_dict):
         rwd_dict['dense'] = np.sum([wt*rwd_dict[key] for key, wt in self.rwd_keys_wt.items()], axis=0)
         return rwd_dict
 
-    def reset(self):
+    def reset(self, **kwargs):
         self.sim.model.site_pos[self.target_sid] = self.np_random.uniform(high=self.target_xyz_range['high'], low=self.target_xyz_range['low'])
         self.sim_obsd.model.site_pos[self.target_sid] = self.sim.model.site_pos[self.target_sid]
-        obs = super().reset(self.init_qpos, self.init_qvel)
+        obs = super().reset(self.init_qpos, self.init_qvel, **kwargs)
         return obs

robohive/envs/arms/reach_base_v0.py

@@ -6,7 +6,7 @@
 ================================================= """
 
 import collections
-import gym
+from robohive.utils import gym
 import numpy as np
 
 from robohive.envs import env_base
@@ -97,8 +97,8 @@ def get_reward_dict(self, obs_dict):
         rwd_dict['dense'] = np.sum([wt*rwd_dict[key] for key, wt in self.rwd_keys_wt.items()], axis=0)
         return rwd_dict
 
-    def reset(self, reset_qpos=None, reset_qvel=None):
+    def reset(self, reset_qpos=None, reset_qvel=None, **kwargs):
         self.sim.model.site_pos[self.target_sid] = self.np_random.uniform(high=self.target_xyz_range['high'], low=self.target_xyz_range['low'])
         self.sim_obsd.model.site_pos[self.target_sid] = self.sim.model.site_pos[self.target_sid]
-        obs = super().reset(reset_qpos, reset_qvel)
+        obs = super().reset(reset_qpos, reset_qvel, **kwargs)
         return obs

robohive/envs/claws/__init__.py

@@ -5,7 +5,7 @@
 License :: Under Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
 ================================================= """
 
-from gym.envs.registration import register
+from robohive.utils import gym; register=gym.register
 import os
 curr_dir = os.path.dirname(os.path.abspath(__file__))
 from robohive.envs.env_variants import register_env_variant
@@ -38,7 +38,7 @@
         'model_path': curr_dir+'/trifinger/trifinger_reorient.xml',
         'object_site_name': "object",
         'target_site_name': "target",
-        'target_xyz_range': {'high':[.05, .05, 0.9], 'low':[-.05, -.05, 0.99]},
+        'target_xyz_range': {'high':[.05, .05, 0.99], 'low':[-.05, -.05, 0.9]},
         'target_euler_range': {'high':[1, 1, 1], 'low':[-1, -1, -1]}
     }
 )

robohive/envs/claws/reorient_v0.py

@@ -6,7 +6,7 @@
 ================================================= """
 
 import collections
-import gym
+from robohive.utils import gym
 import numpy as np
 
 from robohive.envs import env_base
@@ -98,7 +98,7 @@ def get_reward_dict(self, obs_dict):
         rwd_dict['dense'] = np.sum([wt*rwd_dict[key] for key, wt in self.rwd_keys_wt.items()], axis=0)
         return rwd_dict
 
-    def reset(self):
+    def reset(self, **kwargs):
         desired_pos = self.np_random.uniform(high=self.target_xyz_range['high'], low=self.target_xyz_range['low'])
         self.sim.model.site_pos[self.target_sid] = desired_pos
         self.sim_obsd.model.site_pos[self.target_sid] = desired_pos
@@ -108,5 +108,5 @@ def reset(self):
         self.sim.model.site_quat[self.target_sid] = euler2quat(desired_orien)
         self.sim_obsd.model.site_quat[self.target_sid] = euler2quat(desired_orien)
 
-        obs = super().reset(self.init_qpos, self.init_qvel)
+        obs = super().reset(self.init_qpos, self.init_qvel, **kwargs)
         return obs