Real-time 3D Map Generation and Analysis Using Multi-channel LiDAR Sensor

Accurate 3D environment reconstruction or 3D mapping is used in many fields, including robotics, autonomous vehicles, and virtual reality. LiDAR (Light Detection and Ranging) technology has emerged as a powerful sensing modality for capturing detailed 3D information about the surrounding environment. LiDAR comes with multiple channels and different field-of-view. The higher the number of channels denser will be the point cloud data. In this paper, the 3D map of a long stretch inside the IIT Hyderabad campus is created using the Velodyne LiDAR having 16, 32, 64, and 128 channels and their analysis are done. The number of LiDAR channels determines the density and resolution of the acquired point cloud. Higher channel configurations generally result in denser point clouds with finer details. However, dense point clouds can be computationally expensive to process and may require more storage space. Selecting keyframes to construct the 3D map reduces the computational time per frame, and the 3D reconstruction is done in real-time, even with the denser point cloud of 128-channel LiDAR. Lower channel configurations give sparse point cloud data, which may lack distinct features necessary for reliable matching. This paper comprehensively analyzes the 3D map created using Velodyne LiDAR with 16, 32, 64, and 128 channels.