Light detection and ranging (LiDAR) as a significant approach to three-dimensional perception has increasingly suffered from challenges for compact structure, high resolution and strong adaptability. This paper presents a versatile LiDAR system that incorporates multi-beam scanning using Risley prisms and coherent detection based on triangular frequency modulation. By combining Risley-prism-based multi-beam steering model with simultaneous distance and velocity measurements, the LiDAR architecture for three-dimensional imaging is theoretically demonstrated. A unified LiDAR point cloud processing framework is developed, where the multi-mode integrated filtering method is proposed for outlier removal using point cloud characteristics. The coarse-fine coupled strategy is also formulated to combine statistical modeling with iterative optimization for multi-channel point cloud registration. It is experimentally validated that our LiDAR can achieve high-resolution three-dimensional information acquisition against long range, and the proposed point cloud processing technique can fully reconstruct the spatial form of object while preserving sufficient details.
Read full abstract