LiDAR Processing
02
Velodyne VLP-16 Decoder
Real-time UDP packet decoder reconstructing 3D point clouds at 300,000+ points per second — the perception foundation for the AVL autonomous go-kart.
Year
2025 — present
Role
Research Assistant, Autonomous Vehicle Laboratory
Stack
Python · velodyne_decoder · Open3D · NumPy · ROS 2
Technical Highlights
- 01300,000+ points per second real-time processing at a stable 15 FPS, validated against VeloView ground truth
- 02Custom UDP packet decoder parsing raw sensor packets into structured 3D point clouds with height-based coloring
- 03Configurable range and angle filtering; supports live streaming, PCAP replay, and ROS bag playback
- 04Threaded capture/render architecture sustaining throughput across x86 and Jetson AGX Orin hardware
- 05Feeds downstream into occupancy grid mapping, RTAB-Map SLAM, and path planning on the AVL-002 platform