Implementation:Haosulab ManiSkill PullCubeTool

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

Overview

Concrete implementation of the tool-assisted cube pulling task environment in ManiSkill.

Description

The PullCubeToolEnv requires a robot to use an L-shaped tool to pull a cube that is out of the robot's direct reach into a reachable region. The tool is within reach of the robot, but the cube is positioned beyond the arm's reach.

Registered as PullCubeTool-v1 with max_episode_steps=100. Supported robots: panda and fetch. Reward modes include "normalized_dense", "dense", "sparse", and "none".

Randomizations: cube position is in a region out of direct arm reach but within tool reach, and the goal region is within arm reach. The L-shaped tool is placed near the robot. Success requires pulling the cube into the target region using the tool.

Usage

Use this environment for tool-use manipulation research. It tests the ability to understand and leverage tool affordances to extend the robot's effective workspace.

Code Reference

Source Location

• Repository: Haosulab_ManiSkill
• File: mani_skill/envs/tasks/tabletop/pull_cube_tool.py

Signature

@register_env("PullCubeTool-v1", max_episode_steps=100)
class PullCubeToolEnv(BaseEnv):
    SUPPORTED_ROBOTS = ["panda", "fetch"]
    agent: Union[Panda, Fetch]

Import

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

I/O Contract

Inputs

Name	Type	Required	Description
obs_mode	str	No	Observation mode
reward_mode	str	No	Reward mode: "normalized_dense", "dense", "sparse", "none"
control_mode	str	No	Control mode for the robot

Outputs

Name	Type	Description
obs	dict/array	Observation including tool pose, cube pose, goal position, TCP pose
reward	float	Dense reward based on reaching tool, grasping, pulling cube to goal
terminated	bool	Whether episode ended by success
truncated	bool	Whether episode hit max steps (100)
info	dict	Contains success flag

Usage Examples

Basic Usage

import gymnasium as gym
import mani_skill.envs

env = gym.make("PullCubeTool-v1", obs_mode="state", render_mode="rgb_array")
obs, info = env.reset()
for _ in range(100):
    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