Advance robotics reinforcement learning with Isaac Lab on DGX Spark
Introduction
Manipulate objects with a Franka 7-DOF robot arm
Train contact-rich manipulation policies with Isaac Lab on DGX Spark
Train multiple agents to coordinate two Shadow Hands in one simulation
Reproduce natural motion with Adversarial Motion Priors
Choose a reinforcement learning library for your task
Next Steps
Advance robotics reinforcement learning with Isaac Lab on DGX Spark
Introduction
Manipulate objects with a Franka 7-DOF robot arm
Train contact-rich manipulation policies with Isaac Lab on DGX Spark
Train multiple agents to coordinate two Shadow Hands in one simulation
Reproduce natural motion with Adversarial Motion Priors
Choose a reinforcement learning library for your task
Next Steps
Choose your technical toolkit
Isaac Lab integrates several reinforcement learning (RL) libraries.
You’ve used Robotic Systems Lab Reinforcement Learning (RSL-RL) for manipulation, RL Games for factory, and skrl for Multi-Agent Proximal Policy Optimization (MAPPO) and Adversarial Motion Priors (AMP) tasks. Now, you’ll compare why each task uses a different library.
Library choice affects the agent configuration, checkpoint format, and available algorithms. Start with a library that has an upstream agent configuration for your task, then consider algorithm support and workflow needs.
The following table summarizes four commonly used reinforcement learning libraries in Isaac Lab, with links to their repositories and the task profiles they fit best:
| Library | Core strength | Best fit |
|---|---|---|
| RSL-RL | Efficient on-policy training | Locomotion and many manager-based tasks |
| RL Games | Recurrent-policy support | Factory tasks and other registered RL Games configurations |
skrl
| MAPPO, Independent PPO (IPPO), and AMP support | Multi-agent and motion-prior tasks |
| Stable Baselines3 (SB3) | Standardized algorithms and API | When the environment registers an SB3 configuration and supports baselines and prototyping |
No single library is the best choice for every task. A practical strategy is to start with the tool that helps you establish a baseline quickly, then move to a more specialized training stack when the task requires it.
What you’ve accomplished
You’ve progressed from basic manipulation to workflow-level decisions for Isaac Lab on Arm. You’ve practiced task selection, library tradeoffs, MARL and AMP workflows, and when distributed training is worth considering.
Next, you can adapt these scripts as reference implementations for your own USD assets, robot models, scenes, and task constraints. Start with a single-GPU baseline, then expand only when workload scale requires it.