05/2026 – 07/2026
Indian Institute of Science (IISc)
Bangalore, India
Research Intern — Vision-Language-Action (VLA) Robotics
- Deployed and investigated OmniVLA, a pretrained Vision-Language-Action model, on a physical 6 wheel differential drive rover; identified the goal reaching "stopping problem" and designed five OmniVLA native stopping solutions using predicted trajectory outputs.
- Designed a modular multi robot navigation architecture for two autonomous rovers, combining OmniVLA for global direction with ORCA for local collision avoidance, with task scheduling, path planning, and swarm robotics considerations.
- Assembled, wired, and integrated a second rover; deployed ROS 2, OmniVLA, camera streaming, and Zenoh on Raspberry Pi and successfully tested dual rover navigation with simultaneous collision avoidance.
