Add assemble_trocar task for G129-Dex3 and fix reset rendering warmup for RLinf extension

source/isaaclab_contrib/isaaclab_contrib/rl/rlinf/extension.py

@@ -468,6 +468,26 @@ def make_env_isaaclab() -> tuple:
                 isaac_env_cfg.scene.num_envs = self.cfg.init_params.num_envs
 
                 env = gym.make(self.isaaclab_env_id, cfg=isaac_env_cfg, render_mode="rgb_array").unwrapped
+
+                _original_reset = env.reset
+
+                import omni.kit.app
+
+                _app = omni.kit.app.get_app()
+
+                def _patched_reset(*args, **kwargs):
+                    obs, extras = _original_reset(*args, **kwargs)
+                    env.sim.set_setting("/app/player/playSimulations", False)
+                    _app.update()
+                    env.sim.set_setting("/app/player/playSimulations", True)
+                    for sensor in env.scene.sensors.values():
+                        sensor.update(dt=0.0, force_recompute=True)
+                    obs = env.observation_manager.compute(update_history=True)
+                    env.obs_buf = obs
+                    return obs, extras
+
+                env.reset = _patched_reset
+
                 return env, sim_app
 
             return make_env_isaaclab
@@ -481,7 +501,6 @@ def _wrap_obs(self, obs: dict) -> dict:
             - ``"extra_view_images"``: ``(B, N, H, W, C)`` — stacked extra cameras.
             - ``"states"``: ``(B, D)`` — concatenated state vector.
             - ``"task_descriptions"``: ``list[str]`` — task descriptions.
-
             Config is read from the YAML file via :func:`_get_isaaclab_cfg`.
 
             Args:

source/isaaclab_tasks/isaaclab_tasks/manager_based/manipulation/assemble_trocar/__init__.py

@@ -0,0 +1,35 @@
+# Copyright (c) 2022-2026, The Isaac Lab Project Developers (https://github.com/isaac-sim/IsaacLab/blob/main/CONTRIBUTORS.md).
+# All rights reserved.
+#
+# SPDX-License-Identifier: BSD-3-Clause
+
+"""Configuration for Install Trocar task with G129 + Dex3 robot.
+
+This module registers the Install Trocar task in IsaacLab's gymnasium registry,
+allowing it to be discovered and used through IsaacLab's standard task interfaces.
+"""
+
+import gymnasium as gym
+
+##
+# Register Gym environments.
+##
+
+
+gym.register(
+    id="Isaac-Assemble-Trocar-G129-Dex3-v0",
+    entry_point="isaaclab.envs:ManagerBasedRLEnv",
+    kwargs={
+        "env_cfg_entry_point": f"{__name__}.g129_dex3_env_cfg:G1AssembleTrocarEnvCfg",
+    },
+    disable_env_checker=True,
+)
+
+gym.register(
+    id="Isaac-Assemble-Trocar-G129-Dex3-Eval-v0",
+    entry_point="isaaclab.envs:ManagerBasedRLEnv",
+    kwargs={
+        "env_cfg_entry_point": f"{__name__}.g129_dex3_env_cfg:G1AssembleTrocarEvalEnvCfg",
+    },
+    disable_env_checker=True,
+)

source/isaaclab_tasks/isaaclab_tasks/manager_based/manipulation/assemble_trocar/config/__init__.py

@@ -0,0 +1,9 @@
+# Copyright (c) 2022-2026, The Isaac Lab Project Developers (https://github.com/isaac-sim/IsaacLab/blob/main/CONTRIBUTORS.md).
+# All rights reserved.
+#
+# SPDX-License-Identifier: BSD-3-Clause
+
+from .camera_config import CameraBaseCfg, CameraPresets
+from .robot_config import G1RobotPresets
+
+__all__ = ["G1RobotPresets", "CameraBaseCfg", "CameraPresets"]

source/isaaclab_tasks/isaaclab_tasks/manager_based/manipulation/assemble_trocar/config/camera_config.py

@@ -0,0 +1,131 @@
+# Copyright (c) 2022-2026, The Isaac Lab Project Developers (https://github.com/isaac-sim/IsaacLab/blob/main/CONTRIBUTORS.md).
+# All rights reserved.
+#
+# SPDX-License-Identifier: BSD-3-Clause
+
+"""
+public camera configuration
+include the basic configuration for different types of cameras, support scene-specific parameter customization
+"""
+
+from collections.abc import Sequence
+
+import isaaclab.sim as sim_utils
+from isaaclab.sensors import CameraCfg, TiledCameraCfg
+from isaaclab.utils import configclass
+
+
+@configclass
+class CameraBaseCfg:
+    """camera base configuration class
+
+    provide the default configuration for different types of cameras, support scene-specific parameter customization
+    """
+
+    @classmethod
+    def get_camera_config(
+        cls,
+        prim_path: str = "/World/envs/env_.*/Robot/d435_link/front_cam",
+        update_period: float = 0.02,
+        height: int = 480,
+        width: int = 640,
+        focal_length: float = 7.6,
+        focus_distance: float = 400.0,
+        horizontal_aperture: float = 20.0,
+        clipping_range: tuple[float, float] = (0.1, 1.0e5),
+        pos_offset: tuple[float, float, float] = (0.0, 0.0, 0.0),
+        rot_offset: tuple[float, float, float, float] = (0.5, -0.5, 0.5, -0.5),
+        data_types: Sequence[str] | None = None,
+    ) -> CameraCfg:
+        """Get a pinhole camera configuration.
+
+        Args:
+            prim_path: the path of the camera in the scene
+            update_period: update period (seconds)
+            height: image height (pixels)
+            width: image width (pixels)
+            focal_length: focal length
+            focus_distance: focus distance
+            horizontal_aperture: horizontal aperture
+            clipping_range: clipping range (near clipping plane, far clipping plane)
+            pos_offset: position offset (x, y, z)
+            rot_offset: rotation offset quaternion
+            data_types: data type list
+
+        Returns:
+            CameraCfg: camera configuration
+        """
+        if data_types is None:
+            data_types = ("rgb",)
+
+        return TiledCameraCfg(
+            prim_path=prim_path,
+            update_period=update_period,
+            height=height,
+            width=width,
+            data_types=list(data_types),
+            spawn=sim_utils.PinholeCameraCfg(
+                focal_length=focal_length,
+                focus_distance=focus_distance,
+                horizontal_aperture=horizontal_aperture,
+                clipping_range=clipping_range,
+            ),
+            offset=TiledCameraCfg.OffsetCfg(pos=pos_offset, rot=rot_offset, convention="ros"),
+        )
+
+
+@configclass
+class CameraPresets:
+    """camera preset configuration collection
+
+    include the common camera configuration preset for different scenes
+    """
+
+    @classmethod
+    def g1_front_camera(cls, **overrides) -> CameraCfg:
+        params = {
+            "height": 224,
+            "width": 224,
+            "focal_length": 10.5,
+            "horizontal_aperture": 14.25,  # Match original vertical FOV after crop
+        }
+        params.update(overrides)
+        return CameraBaseCfg.get_camera_config(**params)
+
+    @classmethod
+    def left_dex3_wrist_camera(cls, **overrides) -> CameraCfg:
+        """left wrist camera configuration"""
+        params = {
+            "prim_path": "/World/envs/env_.*/Robot/left_hand_camera_base_link/left_wrist_camera",
+            "height": 224,
+            "width": 224,
+            "update_period": 0.02,
+            "data_types": ["rgb"],
+            "focal_length": 12.0,
+            "focus_distance": 400.0,
+            "horizontal_aperture": 14.25,  # Match original vertical FOV after crop
+            "clipping_range": (0.1, 1.0e5),
+            "pos_offset": (-0.04012, -0.07441, 0.15711),
+            "rot_offset": (0.00539, 0.86024, 0.0424, 0.50809),
+        }
+        params.update(overrides)
+        return CameraBaseCfg.get_camera_config(**params)
+
+    @classmethod
+    def right_dex3_wrist_camera(cls, **overrides) -> CameraCfg:
+        """right wrist camera configuration"""
+        params = {
+            "prim_path": "/World/envs/env_.*/Robot/right_hand_camera_base_link/right_wrist_camera",
+            "height": 224,
+            "width": 224,
+            "update_period": 0.02,
+            "data_types": ["rgb"],
+            "focal_length": 12.0,
+            "focus_distance": 400.0,
+            "horizontal_aperture": 14.25,  # Match original vertical FOV after crop
+            "clipping_range": (0.1, 1.0e5),
+            "pos_offset": (-0.04012, 0.07441, 0.15711),
+            "rot_offset": (0.00539, 0.86024, 0.0424, 0.50809),
+        }
+        params.update(overrides)
+        return CameraBaseCfg.get_camera_config(**params)

source/isaaclab_tasks/isaaclab_tasks/manager_based/manipulation/assemble_trocar/config/gr00t_config.py

@@ -0,0 +1,144 @@
+# Copyright (c) 2022-2026, The Isaac Lab Project Developers (https://github.com/isaac-sim/IsaacLab/blob/main/CONTRIBUTORS.md).
+# All rights reserved.
+#
+# SPDX-License-Identifier: BSD-3-Clause
+
+"""GR00T data configuration for IsaacLab tasks.
+
+This module defines customizable GR00T data configurations for different
+embodiments. Users can create their own data config classes by subclassing
+BaseDataConfig or copying/modifying the examples here.
+
+Example usage in run.sh:
+    export RLINF_DATA_CONFIG="policy.gr00t_config"
+    export RLINF_DATA_CONFIG_CLASS="policy.gr00t_config:IsaacLabDataConfig"
+"""
+
+from gr00t.data.dataset import ModalityConfig
+from gr00t.data.transform.base import ComposedModalityTransform
+from gr00t.data.transform.concat import ConcatTransform
+from gr00t.data.transform.state_action import StateActionSinCosTransform, StateActionToTensor, StateActionTransform
+from gr00t.data.transform.video import VideoColorJitter, VideoToNumpy, VideoToTensor
+from gr00t.experiment.data_config import DATA_CONFIG_MAP, BaseDataConfig
+from gr00t.model.transforms import GR00TTransform
+
+
+class IsaacLabDataConfig(BaseDataConfig):
+    """Generic GR00T data config for IsaacLab tasks with G1 + Dex3."""
+
+    # Video modality keys (from gr00t_mapping.video in RLINF_OBS_MAP_JSON)
+    video_keys = [
+        "video.left_wrist_view",
+        "video.right_wrist_view",
+        "video.room_view",
+    ]
+
+    # State modality keys (from gr00t_mapping.state in RLINF_OBS_MAP_JSON)
+    state_keys = [
+        "state.left_arm",
+        "state.right_arm",
+        "state.left_hand",
+        "state.right_hand",
+    ]
+
+    # Action modality keys (output from GR00T model)
+    action_keys = [
+        "action.left_arm",
+        "action.right_arm",
+        "action.left_hand",
+        "action.right_hand",
+    ]
+
+    # Language annotation key
+    language_keys = ["annotation.human.task_description"]
+
+    # Observation and action indices
+    observation_indices = [0]
+    action_indices = list(range(16))
+
+    def modality_config(self) -> dict[str, ModalityConfig]:
+        """Define modality configurations for video, state, action, and language."""
+        video_modality = ModalityConfig(
+            delta_indices=self.observation_indices,
+            modality_keys=self.video_keys,
+        )
+
+        state_modality = ModalityConfig(
+            delta_indices=self.observation_indices,
+            modality_keys=self.state_keys,
+        )
+
+        action_modality = ModalityConfig(
+            delta_indices=self.action_indices,
+            modality_keys=self.action_keys,
+        )
+
+        language_modality = ModalityConfig(
+            delta_indices=self.observation_indices,
+            modality_keys=self.language_keys,
+        )
+
+        return {
+            "video": video_modality,
+            "state": state_modality,
+            "action": action_modality,
+            "language": language_modality,
+        }
+
+    def transform(self):
+        """Define the transform pipeline for processing observations and actions."""
+        transforms = [
+            # Video transforms
+            VideoToTensor(apply_to=self.video_keys),
+            # Disabled: camera already outputs 224×224 via TiledCameraCfg.
+            # To avoid VideoToTensor size-check errors, either:
+            #   1. Disable input size validation in VideoToTensor, OR
+            #   2. Set modality meta height/width to 224 to match actual input.
+            # Re-enable VideoCrop/VideoResize if camera resolution changes.
+            # VideoCrop(apply_to=self.video_keys, scale=0.95),
+            # VideoResize(
+            #     apply_to=self.video_keys,
+            #     height=224,
+            #     width=224,
+            #     interpolation="linear",
+            # ),
+            VideoColorJitter(
+                apply_to=self.video_keys,
+                brightness=0.3,
+                contrast=0.4,
+                saturation=0.5,
+                hue=0.08,
+            ),
+            VideoToNumpy(apply_to=self.video_keys),
+            # State transforms
+            StateActionToTensor(apply_to=self.state_keys),
+            StateActionSinCosTransform(apply_to=self.state_keys),
+            # Action transforms
+            StateActionToTensor(apply_to=self.action_keys),
+            StateActionTransform(
+                apply_to=self.action_keys,
+                normalization_modes={key: "min_max" for key in self.action_keys},
+            ),
+            # Concat transforms
+            ConcatTransform(
+                video_concat_order=self.video_keys,
+                state_concat_order=self.state_keys,
+                action_concat_order=self.action_keys,
+            ),
+            # Model-specific transform
+            GR00TTransform(
+                state_horizon=len(self.observation_indices),
+                action_horizon=len(self.action_indices),
+                max_state_dim=64,
+                max_action_dim=32,
+            ),
+        ]
+        return ComposedModalityTransform(transforms=transforms)
+
+
+# --------------------------------------------------------------------------
+# Register data configs into GR00T's DATA_CONFIG_MAP
+# --------------------------------------------------------------------------
+
+# This allows load_data_config("policy.gr00t_config:IsaacLabDataConfig") to work
+DATA_CONFIG_MAP["isaaclab_g1_dex3"] = IsaacLabDataConfig()