Developed a data pipeline to fine-tune OpenVLA, an open-source vision-language-action model, leveraging IsaacSim for large-scale simulated image collection with randomized camera setups, demonstrating improved model robustness and performance.
No download links available.
