Implementation:Haosulab ManiSkill PegInsertionSide

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

Overview

Concrete implementation of the side peg insertion task environment in ManiSkill.

Description

The PegInsertionSideEnv requires a robot to pick up a peg and insert it sideways into a box with a hole. The box is constructed with four walls forming a square hole, and the peg must be aligned and inserted through the hole along the x-axis.

Registered as PegInsertionSide-v1 with max_episode_steps=100. The supported robot is panda_wristcam (PandaWristCam). Reward modes include "normalized_dense", "dense", "sparse", and "none".

Randomizations include the peg's xy position and z-rotation on the table, and the box's xy position and z-rotation. The peg has configurable half-length (default 0.12) and half-width. Success is determined by the peg being inserted deep enough into the hole.

Usage

Use this environment for precision manipulation research involving alignment and insertion tasks. It tests fine motor control and spatial reasoning abilities.

Code Reference

Source Location

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

Signature

@register_env("PegInsertionSide-v1", max_episode_steps=100)
class PegInsertionSideEnv(BaseEnv):
    SUPPORTED_ROBOTS = ["panda_wristcam"]
    agent: Union[PandaWristCam]
    peg_half_width = 0.025
    peg_half_length = 0.12

Import

import gymnasium as gym
import mani_skill.envs
env = gym.make("PegInsertionSide-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 Panda robot

Outputs

Name	Type	Description
obs	dict/array	Observation including peg pose, box pose, TCP pose
reward	float	Reward based on reaching, grasping, alignment, and insertion
terminated	bool	Whether episode ended by success/failure
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("PegInsertionSide-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