Hey all
Over the past few weeks, I’ve been working on a sim2real pipeline to bring a simple reinforcement learning reach task from simulation to a real Kinova Gen3 arm. I used Isaac Lab for training and deployed everything through ROS 2.
The repo includes:
- RL training scripts using Isaac Lab
- ROS 2-only deployment (no simulator needed at runtime)
- A trained policy you can test right away on hardware
It’s meant to be simple, modular, and a good base for building on. Hope it’s useful or sparks some ideas for others working on sim2real or robotic manipulation!
~ Louis