LiDAR Based Rescue Robot

Our team developed an autonomous vehicle capable of navigating a randomly configured maze and collecting tennis balls, integrating electrical, mechanical, and computer vision components. The robot uses a custom tank chassis for stability and customizability, powered by a 20V battery system with voltage regulators to efficiently manage power distribution. Key components include an Nvidia mini-PC, Intel RealSense camera for depth sensing, and a LiDAR for spatial mapping. The software integrates a pre-trained YOLO (You Only Look Once) model for object detection, combining LiDAR data and depth information to dynamically adapt to the environment. Despite challenges in optimizing wall-following behavior for the competition maze, the project showcased our ability to merge diverse technologies into a functional system, highlighting areas for future improvement in navigation strategies and hardware design.