So-Arm Training using Reinforcement Learning (PPO) in Isaac Lab

Training a robotic arm using Proximal Policy Optimization (PPO) reinforcement learning algorithm in NVIDIA Isaac Lab simulation environment.

This project focuses on training a robotic arm using Reinforcement Learning, specifically the Proximal Policy Optimization (PPO) algorithm, within the NVIDIA Isaac Lab simulation environment.

Overview

The SO-101 Arm RL Controller project demonstrates the application of modern deep reinforcement learning techniques to robotic manipulation tasks. I trained a PPO policy in Isaac Lab with custom reward shaping and am currently implementing deployment on real hardware for multi-pose execution.

Technical Details

  • Algorithm: Proximal Policy Optimization (PPO) with custom reward shaping
  • Simulation Environment: NVIDIA Isaac Lab
  • Task: Robotic arm manipulation and control
  • Status: Currently implementing deployment on real hardware for multi-pose execution
So-Arm training in Isaac Lab environment.

Results