Reinforcement Learning

The following workflows build their environments in Isaac Lab Arena and plug them into Isaac Lab’s reinforcement-learning workflows, covering environment setup, policy training in Isaac Lab, and closed-loop evaluation back in Isaac Lab Arena.

Currently, the following reinforcement learning workflow examples are provided:

• Franka Lift Object Task — Franka Panda, PhysX, joint-position control

• Dexsuite Kuka Allegro Lift Task (Newton) — Kuka + Allegro hand, Newton physics, dexterous manipulation

• Franka Lift Object Task
• Dexsuite Kuka Allegro Lift Task (Newton)