Implementation:Haosulab ManiSkill StackPyramid

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

Overview

Concrete implementation of the pyramid stacking task environment in ManiSkill.

Description

The StackPyramidEnv requires a robot to arrange three cubes into a pyramid: pick up a red cube, place it next to a green cube, then stack a blue cube on top of both.

Registered as StackPyramid-v1 with max_episode_steps=250. Supported robots: panda and fetch. Reward modes include "normalized_dense", "dense", "sparse", and "none".

Randomizations: all three cubes have z-axis rotation randomized, xy positions randomized on the table such that they do not collide. Success requires the pyramid structure to be stable with the blue cube resting on top of the red and green cubes without falling.

Usage

Use this environment for multi-step sequential manipulation research. The pyramid task requires planning and executing a specific sequence of pick-and-place operations.

Code Reference

Source Location

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

Signature

@register_env("StackPyramid-v1", max_episode_steps=250)
class StackPyramidEnv(BaseEnv):
    SUPPORTED_ROBOTS = ["panda", "fetch"]
    agent: Union[Panda, Fetch]

Import

import gymnasium as gym
import mani_skill.envs
env = gym.make("StackPyramid-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, three cube poses, is_grasped
reward	float	Dense reward based on sequential placement milestones
terminated	bool	Whether episode ended by success
truncated	bool	Whether episode hit max steps (250)
info	dict	Contains success flag and stage progress

Usage Examples

Basic Usage

import gymnasium as gym
import mani_skill.envs

env = gym.make("StackPyramid-v1", obs_mode="state", render_mode="rgb_array")
obs, info = env.reset()
for _ in range(250):
    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