Implementation:Haosulab ManiSkill PutOnInScene

Knowledge Sources	Haosulab_ManiSkill
Domains	Robotics, Simulation, Digital_Twin
Last Updated	2026-02-15 08:00 GMT

Overview

Concrete implementation of digital twin pick-and-place evaluation tasks from the Bridge v2 dataset in ManiSkill.

Description

This file defines several digital twin environments that replicate real-world Bridge v2 dataset scenes for sim-to-real evaluation. All classes extend BaseBridgeEnv. The registered environments include:

• PutCarrotOnPlateInScene-v1 (max_episode_steps=60): Put a carrot on a plate on a flat table.
• PutEggplantInBasketScene-v1 (max_episode_steps=120): Put an eggplant into a yellow basket in a sink scene.
• StackGreenCubeOnYellowCubeBakedTexInScene-v1 (max_episode_steps=60): Stack a green cube on a yellow cube with baked textures.
• PutSpoonOnTableClothInScene-v1 (max_episode_steps=60): Put a spoon on a towel/table cloth.

All variants require asset_download_ids=["bridge_v2_real2sim"]. Each environment defines specific object configurations (xyz_configs, quat_configs) and evaluation criteria. Language instructions are provided via get_language_instruction().

Usage

Use these environments for sim-to-real evaluation of manipulation policies trained on the Bridge v2 dataset. They provide realistic digital twin scenes matching real-world setups.

Code Reference

Source Location

• Repository: Haosulab_ManiSkill
• File: mani_skill/envs/tasks/digital_twins/bridge_dataset_eval/put_on_in_scene.py

Signature

@register_env(
    "PutCarrotOnPlateInScene-v1",
    max_episode_steps=60,
    asset_download_ids=["bridge_v2_real2sim"],
)
class PutCarrotOnPlateInScene(BaseBridgeEnv): ...

@register_env(
    "PutEggplantInBasketScene-v1",
    max_episode_steps=120,
    asset_download_ids=["bridge_v2_real2sim"],
)
class PutEggplantInBasketScene(BaseBridgeEnv): ...

Import

import gymnasium as gym
import mani_skill.envs
env = gym.make("PutCarrotOnPlateInScene-v1")

I/O Contract

Inputs

Name	Type	Required	Description
obs_mode	str	No	Observation mode
reward_mode	str	No	Reward mode
control_mode	str	No	Control mode

Outputs

Name	Type	Description
obs	dict/array	Observation based on obs_mode
reward	float	Task reward
terminated	bool	Whether episode ended by success/failure
truncated	bool	Whether episode hit max steps
info	dict	Contains success flag

Usage Examples

Basic Usage

import gymnasium as gym
import mani_skill.envs

env = gym.make("PutCarrotOnPlateInScene-v1", obs_mode="state", render_mode="rgb_array")
obs, info = env.reset()
for _ in range(60):
    action = env.action_space.sample()
    obs, reward, terminated, truncated, info = env.step(action)
    if terminated or truncated:
        obs, info = env.reset()
env.close()

Related Pages

• Environment:Haosulab_ManiSkill_Python_SAPIEN_Core