Implementation:Haosulab ManiSkill PickSingleYCB

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

Overview

Concrete implementation of the single YCB object picking task environment in ManiSkill.

Description

The PickSingleYCBEnv picks up a random object from the YCB dataset and moves it to a random goal position. The object geometry is randomized by sampling from the YCB dataset (excluding non-graspable objects like windex_bottle, skillet_lid, plate, and chain).

Registered as PickSingleYCB-v1 with max_episode_steps=50 and asset_download_ids=["ycb"]. Supported robots: panda, panda_wristcam, fetch. Reward modes include "normalized_dense", "dense", "sparse", and "none".

Randomizations: object xy position in [-0.1, 0.1], z-axis rotation, goal position xy in [-0.1, 0.1] with z in [0, 0.3] above the object. Success requires the object to be within 0.025m of the goal and the robot to be static. For GPU simulation, 128+ parallel environments are recommended to sample all YCB objects.

Usage

Use this environment for generalized grasping research with diverse object geometries. The YCB dataset provides realistic objects with varying shapes and sizes.

Code Reference

Source Location

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

Signature

@register_env("PickSingleYCB-v1", max_episode_steps=50, asset_download_ids=["ycb"])
class PickSingleYCBEnv(BaseEnv):
    SUPPORTED_ROBOTS = ["panda", "panda_wristcam", "fetch"]
    agent: Union[Panda, PandaWristCam, Fetch]
    goal_thresh = 0.025

Import

import gymnasium as gym
import mani_skill.envs
env = gym.make("PickSingleYCB-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 TCP pose, object pose, goal position, is_grasped
reward	float	Dense reward: reaching + grasping + placing + static
terminated	bool	Whether episode ended by success
truncated	bool	Whether episode hit max steps (50)
info	dict	Contains success, is_obj_placed, is_robot_static, is_grasped

Usage Examples

Basic Usage

import gymnasium as gym
import mani_skill.envs

env = gym.make("PickSingleYCB-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