Implementation:Haosulab ManiSkill Hopper

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

Overview

Concrete implementation of the Hopper locomotion control task environment in ManiSkill, adapted from the DeepMind Control Suite.

Description

The Hopper environment implements a planar one-legged hopping robot that must learn to stand or hop. The robot is defined inline via MJCF as HopperRobot with joints for hip, knee, and waist, plus planar tracking joints (rootx, rooty, rootz). It uses a PlanarSceneBuilder.

Registered variants:

• MS-HopperStand-v1 (max_episode_steps=200): Stand upright at a height of 0.6m.
• MS-HopperHop-v1 (max_episode_steps=200): Hop forward at a target speed of 2 m/s while maintaining height.

Rewards are composed of standing reward (based on torso height) and hopping reward (based on forward speed). Constants: _STAND_HEIGHT=0.6, _HOP_SPEED=2. Reward modes include "dense", "normalized_dense", and "none".

Usage

Use this environment for benchmarking locomotion control algorithms on a simpler one-legged robot. The standing variant tests balance while the hopping variant tests dynamic locomotion.

Code Reference

Source Location

• Repository: Haosulab_ManiSkill
• File: mani_skill/envs/tasks/control/hopper.py

Signature

@register_env("MS-HopperStand-v1", max_episode_steps=200)
class HopperStandEnv(HopperEnvBase): ...

@register_env("MS-HopperHop-v1", max_episode_steps=200)
class HopperHopEnv(HopperEnvBase): ...

Import

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

I/O Contract

Inputs

Name	Type	Required	Description
obs_mode	str	No	Observation mode (default: "state")
reward_mode	str	No	Reward mode: "dense", "normalized_dense", "none"
control_mode	str	No	Control mode for the hopper joints

Outputs

Name	Type	Description
obs	dict/array	Observation including joint angles, velocities, torso height
reward	float	Reward based on standing height and forward speed
terminated	bool	Whether episode ended
truncated	bool	Whether episode hit max steps (200)
info	dict	Contains evaluation metrics

Usage Examples

Basic Usage

import gymnasium as gym
import mani_skill.envs

env = gym.make("MS-HopperStand-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