GUARD.IAM is a fully autonomous rover with dual collaborative robotic arms, built for operating in hazardous environments where putting a person on site is risky. When a task needs human judgment, an operator can seamlessly take over through a VR headset for natural, first-person teleoperation. My team designed and built the entire system from scratch in a single 36-hour sprint at Purdue StarkHacks 2026.

It is a full-stack physical-AI platform. We deployed a compact Vision-Language-Action (VLA) model on an AMD Ryzen AI Mini PC for on-board decision making, wrapped a complete ROS2 navigation stack around it, and engineered the custom mechanical hardware that carries the arms, sensors, and compute.

The work earned 2nd Place in the Microsoft Track and 3rd Place for Best Use of AMD Technology, a double win shared with my teammates Phillipp Gery, Vedant Patkar, and Dipam Patel. Thanks to Prof. Yu She and Prof. Shaoshuai Mou of the Purdue Institute for Control, Optimization and Networks for lending the hardware and visiting us on site, to AMD for the edge-computing hardware, and to Microsoft for recognizing the approach.