ppo-pusher-v5 / README.md

huggsook

Upload folder using huggingface_hub

479df53 verified 8 days ago

preview code

Raw

History Blame Contribute Delete

1.1 kB

metadata

title: PPO Pusher-v5 3D Visualizer & RL Simulation
emoji: 🦾
colorFrom: cyan
colorTo: purple
sdk: static
pinned: false
license: mit
tags:
  - reinforcement-learning
  - stable-baselines3
  - ppo
  - gymnasium
  - mujoco
  - threejs
  - simulation

🤖 PPO Pusher-v5 3D Interactive RL Simulation & Visualizer

A complete web-based interactive 3D simulation and learning visualization tool for the Gymnasium Pusher-v5 robotic arm environment trained with Stable-Baselines3 PPO.

✨ Features

3D Robot Arm Simulation (Three.js): Real-time Inverse Kinematics, physics collision, and trajectory visualization.
Interactive Manipulation: Drag the Goal or Puck with your mouse to test the agent's real-time adaptability.
Checkpoint Time Machine: Switch between Step 0 (Untrained), Step 10k (Novice), Step 25k (Intermediate), and Step 50k (Master).
Live Telemetry & Dashboard (Chart.js): Reward convergence curve, policy & value loss dynamics.
One-Click Bundle Export: Download the trained model weights, PyTorch training scripts, and telemetry logs as a .zip file.