Building a Healthcare Robot from Simulation to Deployment with NVIDIA Isaac

A new workflow enables building a surgical assistant robot using NVIDIA Isaac for Healthcare and the LeRobot framework. The process utilizes a Sim2Real mixed training approach, combining synthetic data from simulation with real-world teleoperation data for more robust policies. This end-to-end pipeline covers data collection, fine-tuning the GR00T N1.5 model, and deploying the resulting policy on physical SO-ARM hardware. The guide provides technical implementation details, hardware requirements, and code snippets for the entire process, from data generation to model training and deployment.