Implementation:Haosulab ManiSkill RollBall

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

Overview

Concrete implementation of the ball rolling task environment in ManiSkill.

Description

The RollBallEnv requires a robot to push and roll a ball to a goal region at the other end of the table. The goal is marked by a red/white circular target.

Registered as RollBall-v1 with max_episode_steps=80. Supported robots: panda and fetch. Reward modes include "normalized_dense", "dense", "sparse", and "none".

Randomizations: ball xy position in [0.2, 0.5] x [-0.4, 0.7], goal position in [-0.4, -0.7] x [0.2, -0.9]. Success requires the ball to be within the goal region. This is a non-prehensile manipulation task requiring controlled pushing and rolling of a sphere.

Usage

Use this environment for non-prehensile manipulation research involving rolling contact dynamics. The ball's tendency to roll makes precise control challenging.

Code Reference

Source Location

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

Signature

@register_env("RollBall-v1", max_episode_steps=80)
class RollBallEnv(BaseEnv):
    SUPPORTED_ROBOTS = ["panda", "fetch"]
    agent: Union[Panda, Fetch]

Import

import gymnasium as gym
import mani_skill.envs
env = gym.make("RollBall-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 ball pose, goal position, TCP pose
reward	float	Dense reward based on pushing ball toward goal region
terminated	bool	Whether episode ended by success
truncated	bool	Whether episode hit max steps (80)
info	dict	Contains success flag

Usage Examples

Basic Usage

import gymnasium as gym
import mani_skill.envs

env = gym.make("RollBall-v1", obs_mode="state", render_mode="rgb_array")
obs, info = env.reset()
for _ in range(80):
    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