3D Flash LiDAR Object Detection and Tracking on Edge Hardware

We developed a real-time vehicle detection system combining YOLOX and Kalman filter tracking, optimized for 3D Light detection and ranging (LiDAR) data. Using a TigerCub Flash LiDAR camera, our method eliminates color variability, enhancing detection accuracy. The machine learning models were deployed on an NVIDIA Jetson Orin Nano, achieving over 95% precision in real-time at 30 frames-per-second.