vs050-mujoco 1.1.0

Gymnasium environment for VS050 pick-and-place tasks.

VS050 MuJoCo (Gymnasium)

A complete Gym environment and simulation suite for the DENSO VS050 6-DoF robotic arm extended with a Robotiq 2F-85 intelligent gripper. This repository offers a clean Python package and Gymnasium environments simulating precise manipulation physics via DeepMind MuJoCo.

Installation

This project is fully managed with uv. To install it natively or use it in other projects:

git clone https://github.com/goncamateus/vs050_mujoco.git
cd vs050_mujoco

# Install the package seamlessly into the workspace
uv sync

Quick Start

import gymnasium as gym
import vs050_mujoco  # Registers the environment automatically

env = gym.make("VS050-ReachPose-v0", render_mode="human")
# or
env = gym.make("VS050-PickAndPlace-v0", render_mode="human")
obs, info = env.reset()

for _ in range(1000):
    action = env.action_space.sample()  # Your agent code here
    obs, reward, terminated, truncated, info = env.step(action)
    
    if terminated or truncated:
        obs, info = env.reset()

env.close()

Documentation

The project documentation is cleanly split into two modules covering the simulation physics and the reinforcement learning interface:

• 🤖 Models Documentation: Kinematics, joint limits, meshes, and standalone MuJoCo models.
• 🎯 Environments Documentation: Details on observation spaces, action constraints, and reward functions (includes visual previews).

License

This project integrates assets from multiple open-source sources, distributed under their respective licenses:

• DENSO VS050 Robot Assets (Meshes & XML): Sourced from DENSORobot/denso_robot_ros2 under the MIT and BSD 3-Clause Licenses.
• Robotiq 2F-85 Gripper Assets: Sourced from Google DeepMind's mujoco_menagerie under the Apache License 2.0.
• Environment & Implementation Code: Provided under the MIT License.

See the LICENSE file for the exact terms and complete license texts.