Creating a Modular Environment

Creating a Modular Environment#

This tutorial demonstrates how to create sophisticated robotic environments using EmbodiChain’s modular architecture. You’ll learn how to use the advanced envs.EmbodiedEnv class with configuration-driven setup, event managers, observation managers, and randomization systems.

The Code#

The tutorial corresponds to the modular_env.py script in the scripts/tutorials/gym directory.

Code for modular_env.py
  1# ----------------------------------------------------------------------------
  2# Copyright (c) 2021-2026 DexForce Technology Co., Ltd.
  3#
  4# Licensed under the Apache License, Version 2.0 (the "License");
  5# you may not use this file except in compliance with the License.
  6# You may obtain a copy of the License at
  7#
  8#     http://www.apache.org/licenses/LICENSE-2.0
  9#
 10# Unless required by applicable law or agreed to in writing, software
 11# distributed under the License is distributed on an "AS IS" BASIS,
 12# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 13# See the License for the specific language governing permissions and
 14# limitations under the License.
 15# ----------------------------------------------------------------------------
 16
 17import torch
 18
 19from typing import List, Dict, Any
 20
 21import embodichain.lab.gym.envs.managers.randomization as rand
 22import embodichain.lab.gym.envs.managers.events as events
 23import embodichain.lab.gym.envs.managers.observations as obs
 24
 25from embodichain.lab.gym.envs.managers import (
 26    EventCfg,
 27    SceneEntityCfg,
 28    ObservationCfg,
 29)
 30from embodichain.lab.gym.envs import EmbodiedEnv, EmbodiedEnvCfg
 31from embodichain.lab.gym.utils.registration import register_env
 32from embodichain.lab.sim.robots import DexforceW1Cfg
 33from embodichain.lab.sim.sensors import StereoCameraCfg, SensorCfg
 34from embodichain.lab.sim.shapes import MeshCfg
 35from embodichain.lab.sim.cfg import (
 36    RenderCfg,
 37    LightCfg,
 38    ArticulationCfg,
 39    RobotCfg,
 40    RigidObjectCfg,
 41    RigidBodyAttributesCfg,
 42)
 43from embodichain.data import get_data_path
 44from embodichain.utils import configclass
 45
 46
 47@configclass
 48class ExampleEventCfg:
 49
 50    replace_obj: EventCfg = EventCfg(
 51        func=events.replace_assets_from_group,
 52        mode="reset",
 53        params={
 54            "entity_cfg": SceneEntityCfg(
 55                uid="fork",
 56            ),
 57            "folder_path": get_data_path("TableWare/tableware/fork/"),
 58        },
 59    )
 60
 61    randomize_fork_mass: EventCfg = EventCfg(
 62        func=rand.randomize_rigid_object_mass,
 63        mode="reset",
 64        params={
 65            "entity_cfg": SceneEntityCfg(
 66                uid="fork",
 67            ),
 68            "mass_range": (0.1, 2.0),
 69        },
 70    )
 71
 72    randomize_table_mat: EventCfg = EventCfg(
 73        func=rand.randomize_visual_material,
 74        mode="interval",
 75        interval_step=25,
 76        params={
 77            "entity_cfg": SceneEntityCfg(
 78                uid="table",
 79            ),
 80            "random_texture_prob": 0.5,
 81            "texture_path": get_data_path("CocoBackground/coco"),
 82            "base_color_range": [[0.2, 0.2, 0.2], [1.0, 1.0, 1.0]],
 83        },
 84    )
 85
 86
 87@configclass
 88class ObsCfg:
 89
 90    obj_pose: ObservationCfg = ObservationCfg(
 91        func=obs.get_rigid_object_pose,
 92        mode="add",
 93        name="fork_pose",
 94        params={"entity_cfg": SceneEntityCfg(uid="fork")},
 95    )
 96
 97
 98@configclass
 99class ExampleCfg(EmbodiedEnvCfg):
100
101    # Define the robot configuration using DexforceW1Cfg
102    robot: RobotCfg = DexforceW1Cfg.from_dict(
103        {
104            "uid": "dexforce_w1",
105            "version": "v021",
106            "arm_kind": "anthropomorphic",
107            "init_pos": [0.0, 0, 0.0],
108        }
109    )
110
111    # Define the sensor configuration using StereoCameraCfg
112    sensor: List[SensorCfg] = [
113        StereoCameraCfg(
114            uid="eye_in_head",
115            width=960,
116            height=540,
117            enable_mask=True,
118            enable_depth=True,
119            left_to_right_pos=(0.06, 0, 0),
120            intrinsics=(450, 450, 480, 270),
121            intrinsics_right=(450, 450, 480, 270),
122            extrinsics=StereoCameraCfg.ExtrinsicsCfg(
123                parent="eyes",
124            ),
125        )
126    ]
127
128    background: List[RigidObjectCfg] = [
129        RigidObjectCfg(
130            uid="table",
131            shape=MeshCfg(
132                fpath=get_data_path("CircleTableSimple/circle_table_simple.ply"),
133                compute_uv=True,
134            ),
135            attrs=RigidBodyAttributesCfg(
136                mass=10.0,
137                static_friction=0.95,
138                dynamic_friction=0.85,
139                restitution=0.01,
140            ),
141            body_type="kinematic",
142            init_pos=(0.80, 0, 0.8),
143            init_rot=(0, 90, 0),
144        ),
145    ]
146
147    rigid_object: List[RigidObjectCfg] = [
148        RigidObjectCfg(
149            uid="fork",
150            shape=MeshCfg(
151                fpath=get_data_path("TableWare/tableware/fork/standard_fork_scale.ply"),
152            ),
153            body_scale=(0.75, 0.75, 1.0),
154            init_pos=(0.8, 0, 1.0),
155        ),
156    ]
157
158    articulation_cfg: List[ArticulationCfg] = [
159        ArticulationCfg(
160            uid="drawer",
161            fpath="SlidingBoxDrawer/SlidingBoxDrawer.urdf",
162            init_pos=(0.5, 0.0, 0.85),
163        )
164    ]
165
166    events = ExampleEventCfg()
167
168    observations = ObsCfg()
169
170
171@register_env("ModularEnv-v1", max_episode_steps=100, override=True)
172class ModularEnv(EmbodiedEnv):
173    """
174    An example of a modular environment that inherits from EmbodiedEnv
175    and uses custom event and observation managers.
176    """
177
178    def __init__(self, cfg: EmbodiedEnvCfg, **kwargs):
179        super().__init__(cfg, **kwargs)
180
181
182if __name__ == "__main__":
183    import gymnasium as gym
184    import argparse
185
186    from embodichain.lab.sim import SimulationManagerCfg
187    from embodichain.lab.gym.utils.gym_utils import add_env_launcher_args_to_parser
188
189    parser = argparse.ArgumentParser()
190    add_env_launcher_args_to_parser(parser)
191    args = parser.parse_args()
192
193    env_cfg = ExampleCfg(
194        sim_cfg=SimulationManagerCfg(
195            render_cfg=RenderCfg(renderer=args.renderer),
196            headless=args.headless,
197            sim_device=args.device,
198        ),
199        num_envs=args.num_envs,
200    )
201
202    # Create the Gym environment
203    env = gym.make("ModularEnv-v1", cfg=env_cfg)
204
205    for i in range(5):
206        obs, info = env.reset()
207
208        for i in range(100):
209            action = torch.zeros(env.action_space.shape, dtype=torch.float32)
210            obs, reward, done, truncated, info = env.step(action)

The Code Explained#

This tutorial showcases EmbodiChain’s most powerful environment creation approach using the envs.EmbodiedEnv class. Unlike the basic environment tutorial, this approach uses declarative configuration classes and manager systems for maximum flexibility and reusability.

Event Configuration#

Events define automated behaviors that occur during simulation. There are three types of supported modes:

  • startup: triggers once when the environment is initialized

  • reset: triggers every time the environment is reset

  • interval: triggers at fixed step intervals during simulation

The ExampleEventCfg demonstrates three types of events:

    RenderCfg,
    LightCfg,
    ArticulationCfg,
    RobotCfg,
    RigidObjectCfg,
    RigidBodyAttributesCfg,
)
from embodichain.data import get_data_path
from embodichain.utils import configclass


@configclass
class ExampleEventCfg:

    replace_obj: EventCfg = EventCfg(
        func=events.replace_assets_from_group,
        mode="reset",
        params={
            "entity_cfg": SceneEntityCfg(
                uid="fork",
            ),
            "folder_path": get_data_path("TableWare/tableware/fork/"),
        },
    )

    randomize_fork_mass: EventCfg = EventCfg(
        func=rand.randomize_rigid_object_mass,
        mode="reset",
        params={
            "entity_cfg": SceneEntityCfg(
                uid="fork",
            ),
            "mass_range": (0.1, 2.0),
        },
    )

    randomize_table_mat: EventCfg = EventCfg(
        func=rand.randomize_visual_material,
        mode="interval",
        interval_step=25,
        params={

Asset Replacement Event

The replace_obj event demonstrates dynamic asset swapping:

Light Randomization Event

The randomize_light event creates dynamic lighting conditions:

  • Function: envs.managers.randomization.rendering.randomize_light()

  • Mode: "interval" - triggers every 5 steps

  • Parameters: Randomizes position, color, and intensity within specified ranges

Material Randomization Event

The randomize_table_mat event varies visual appearance:

  • Function: envs.managers.randomization.rendering.randomize_visual_material()

  • Mode: "interval" - triggers every 10 steps

  • Features: Random textures from COCO dataset and base color variations

For more randomization events, please refer to Event Functors.

Observation Configuration#

The default observation from envs.EmbodiedEnv includes: - robot: robot proprioceptive data (joint positions, velocities, efforts) - sensor: all available sensor data (images, depth, segmentation, etc.)

However, users always need to define some custom observation for specified learning tasks. To handle this, the observation manager system allows users to declaratively specify additional observations.

            ),
            "random_texture_prob": 0.5,
            "texture_path": get_data_path("CocoBackground/coco"),
            "base_color_range": [[0.2, 0.2, 0.2], [1.0, 1.0, 1.0]],
        },
    )


@configclass

This configuration:

For details documentation, see envs.managers.cfg.ObservationCfg.

Environment Configuration#

The main environment configuration inherits from envs.EmbodiedEnvCfg and defines all scene components:

Robot Configuration

    robot: RobotCfg = DexforceW1Cfg.from_dict(
        {
            "uid": "dexforce_w1",
            "version": "v021",
            "arm_kind": "anthropomorphic",
            "init_pos": [0.0, 0, 0.0],
        }
    )

Uses the pre-configured DexforceW1Cfg with customizations:

  • Version: Specific robot variant (v021)

  • Arm Type: Anthropomorphic configuration

  • Position: Initial placement in the scene

Sensor Configuration

            "uid": "dexforce_w1",
            "version": "v021",
            "arm_kind": "anthropomorphic",
            "init_pos": [0.0, 0, 0.0],
        }
    )

    # Define the sensor configuration using StereoCameraCfg
    sensor: List[SensorCfg] = [
        StereoCameraCfg(
            uid="eye_in_head",
            width=960,
            height=540,
            enable_mask=True,
            enable_depth=True,

Configures a stereo camera system using StereoCameraCfg:

  • Resolution: 960x540 pixels for realistic visual input

  • Features: Depth sensing and segmentation masks enabled

  • Stereo Setup: 6cm baseline between left and right cameras

  • Mounting: Attached to robot’s “eyes” frame

Lighting Configuration

            intrinsics=(450, 450, 480, 270),
            intrinsics_right=(450, 450, 480, 270),
            extrinsics=StereoCameraCfg.ExtrinsicsCfg(
                parent="eyes",
            ),
        )
    ]

    background: List[RigidObjectCfg] = [
        RigidObjectCfg(
            uid="table",

Defines scene illumination with configurable lights:

  • Types: Supports "point", "sun", "direction", "spot", "rect", and "mesh".

  • Global lights: "sun" and "direction" are global scene lights (single instance, infinite distance).

  • Properties: Configurable color, intensity, position, and direction.

  • UID: Named reference for event system manipulation.

Rigid Objects

                fpath=get_data_path("CircleTableSimple/circle_table_simple.ply"),
                compute_uv=True,
            ),
            attrs=RigidBodyAttributesCfg(
                mass=10.0,
                static_friction=0.95,
                dynamic_friction=0.85,
                restitution=0.01,
            ),
            body_type="kinematic",
            init_pos=(0.80, 0, 0.8),
            init_rot=(0, 90, 0),
        ),
    ]

    rigid_object: List[RigidObjectCfg] = [
        RigidObjectCfg(
            uid="fork",
            shape=MeshCfg(
                fpath=get_data_path("TableWare/tableware/fork/standard_fork_scale.ply"),
            ),
            body_scale=(0.75, 0.75, 1.0),
            init_pos=(0.8, 0, 1.0),
        ),
    ]

Multiple objects demonstrate different physics properties:

Table Configuration:

  • Shape: Custom PLY mesh with UV mapping

  • Physics: Kinematic body (movable but not affected by forces)

  • Material: Friction and restitution properties for realistic contact

Fork Configuration:

  • Shape: Detailed mesh from asset library

  • Scale: Proportionally scaled for scene consistency

  • Physics: Dynamic body affected by gravity and collisions

Articulated Objects

        ArticulationCfg(
            uid="drawer",
            fpath="SlidingBoxDrawer/SlidingBoxDrawer.urdf",
            init_pos=(0.5, 0.0, 0.85),
        )
    ]

    events = ExampleEventCfg()

    observations = ObsCfg()

Demonstrates complex mechanisms with moving parts:

  • URDF: Sliding drawer with joints and constraints

  • Positioning: Placed on table surface for interaction

Environment Implementation#

The actual environment class is remarkably simple due to the configuration-driven approach:

The envs.EmbodiedEnv base class automatically:

  • Loads all configured scene components

  • Sets up observation and action spaces

  • Initializes event and observation managers

  • Handles environment lifecycle (reset, step, etc.)

The Code Execution#

To run the modular environment:

cd /path/to/embodichain
python scripts/tutorials/gym/modular_env.py

The script demonstrates the complete workflow:

  1. Configuration: Creates an instance of ExampleCfg

  2. Registration: Uses the registered environment ID

  3. Execution: Runs episodes with zero actions to observe automatic behaviors

Manager System Benefits#

The manager-based architecture provides several key advantages:

Event Managers

  • Modularity: Reusable event functions across environments

  • Timing Control: Flexible scheduling (reset, interval, condition-based)

  • Parameter Binding: Type-safe configuration with validation

  • Extensibility: Easy to add custom event behaviors

Observation Managers

  • Flexible Data: Any simulation data can become an observation

  • Processing Pipeline: Built-in normalization and transformation

  • Dynamic Composition: Runtime observation space modification

  • Performance: Efficient data collection and GPU acceleration

Key Features Demonstrated#

This tutorial showcases the most advanced features of EmbodiChain environments:

  1. Configuration-Driven Design: Declarative environment specification

  2. Manager Systems: Modular event and observation handling

  3. Asset Management: Dynamic loading and randomization

  4. Sensor Integration: Realistic camera systems with stereo vision

  5. Physics Simulation: Complex articulated and rigid body dynamics

  6. Visual Randomization: Automated domain randomization

  7. Extensible Architecture: Easy customization and extension points

This tutorial demonstrates the full power of EmbodiChain’s modular environment system, providing the foundation for creating sophisticated robotic learning scenarios.

Tip

Using an AI coding agent? These skills can help you build on this tutorial:

  • /add-task-env — Scaffold a new task environment with the correct file structure, @register_env decorator, base class methods, __init__.py update, and test stub.

  • /add-functor — Add observation, reward, event, or randomization functors with the correct signature and module placement.