Adaptive Real-Time Lidar Raw Packet Processing for Dynamic Environments

This paper proposes a system for real-time terrestrial 3D LiDAR data capturing and processing using a 32channel LiDAR sensor. The system supports two methods for data acquisition: a basic approach that outputs packet capture (pcap) files and a custom algorithm designed to store raw LiDAR data directly in binary (bin) and point cloud (pcd) formats. Our pipeline includes continuous frame acquisition via UDP port, enabling accurate visualization in both 2D and 3D formats to ensure the reliability of the captured data. We employ azimuthal and elevation-based calibration for precise calculation of 3D coordinates for each LiDAR scan point. A point cloud generation algorithm efficiently parses, transforms, and stores LiDAR data for applications such as 3D mapping, navigation, and object detection in autonomous systems. Our results confirm that the proposed method provides an effective framework for real-time 3D point cloud generation and validation, significantly enhancing the performance of robotics and autonomous vehicle applications.