Implementation:Haosulab ManiSkill PokeCube

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

Overview

Concrete implementation of the cube poking task environment in ManiSkill.

Description

The PokeCubeEnv requires a robot to poke a red cube with a peg and push it to a target goal position marked by a red/white circular target. The peg is placed flat on the table and the cube is positioned in front of it.

Registered as PokeCube-v1 with max_episode_steps=50. Supported robots: panda and fetch. Reward modes include "normalized_dense", "dense", "sparse", and "none".

Randomizations: peg xy position in [-0.1, 0.1], cube x fixed relative to peg (peg_x + 0.12 + 0.1), cube y in [-0.1, 0.1], cube z-rotation in [-pi/6, pi/6], goal position fixed at cube_xy + [0.05 + goal_radius, 0]. Success requires the cube to be within the goal radius of the target position.

Usage

Use this environment for non-prehensile tool-use manipulation research. The robot must use a peg as an intermediary to push the cube, testing tool-based interaction skills.

Code Reference

Source Location

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

Signature

@register_env("PokeCube-v1", max_episode_steps=50)
class PokeCubeEnv(BaseEnv):
    SUPPORTED_ROBOTS = ["panda", "fetch"]
    agent: Union[Panda, Fetch]

Import

import gymnasium as gym
import mani_skill.envs
env = gym.make("PokeCube-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 peg pose, cube pose, goal position, TCP pose
reward	float	Dense reward based on reaching peg, pushing cube to goal
terminated	bool	Whether episode ended by success
truncated	bool	Whether episode hit max steps (50)
info	dict	Contains success flag

Usage Examples

Basic Usage

import gymnasium as gym
import mani_skill.envs

env = gym.make("PokeCube-v1", obs_mode="state", render_mode="rgb_array")
obs, info = env.reset()
for _ in range(50):
    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