Implementation:Haosulab ManiSkill DrawSVG

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

Overview

Concrete implementation of the SVG drawing task environment in ManiSkill.

Description

The DrawSVGEnv instantiates a table with a white canvas and a robot with a stick that must trace an SVG path. The robot draws red lines by placing colored dots on the canvas when the brush is close enough to the surface. Key parameters include MAX_DOTS=1000 for total ink available, BRUSH_RADIUS=0.01, and a success THRESHOLD of 0.1.

Registered as DrawSVG-v1 with max_episode_steps=500. The supported robot is panda_stick (PandaStick). The reward mode is "sparse" and success is measured by how well drawn points match the goal SVG path within a euclidean distance threshold.

Randomizations include the SVG position on the xy-plane and its z-rotation in range [0, 2*pi].

Usage

Use this environment for learning path-following and drawing behaviors with a robot manipulator. It extends the base drawing environment with SVG path targets.

Code Reference

Source Location

• Repository: Haosulab_ManiSkill
• File: mani_skill/envs/tasks/drawing/draw_svg.py

Signature

@register_env("DrawSVG-v1", max_episode_steps=500)
class DrawSVGEnv(BaseEnv):
    SUPPORTED_REWARD_MODES = ["sparse"]
    SUPPORTED_ROBOTS: ["panda_stick"]
    agent: PandaStick
    MAX_DOTS = 1000
    BRUSH_RADIUS = 0.01
    THRESHOLD = 0.1

Import

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

I/O Contract

Inputs

Name	Type	Required	Description
obs_mode	str	No	Observation mode
reward_mode	str	No	Reward mode: "sparse"
control_mode	str	No	Control mode for panda_stick robot

Outputs

Name	Type	Description
obs	dict/array	Observation including TCP pose and canvas state
reward	float	Sparse reward based on coverage of SVG path
terminated	bool	Whether episode ended by success/failure
truncated	bool	Whether episode hit max steps (500)
info	dict	Contains success flag and coverage metrics

Usage Examples

Basic Usage

import gymnasium as gym
import mani_skill.envs

env = gym.make("DrawSVG-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