Capstone Project – Vision-Based Autonomous Quadcopter

Built an autonomous indoor quadcopter system that uses real-time onboard vision (MobileNet-SSD on Jetson Nano), visual-servoing control, and a ROS 2 state machine to track and follow a fast-moving “getaway car” in a simulated police chase scenario. The project integrates perception, control, and high-level behaviour into a system that can detect, chase, and reacquire a target under tight compute constraints.