Accelerating real-time LiDAR data processing using GPUs

Light Detection and Ranging (LiDAR) sensors are used for acquiring high density topographical data with extremely high spatial resolution. Many LiDAR-based applications, e.g. unmanned autonomous ground and air vehicles require realtime processing capabilities for navigation. The processing of the massive LiDAR data is time consuming due to the magnitude of the data produced and also due to the computationally iterative nature of the algorithms. Graphics Processing Units (GPU) consist of massively parallel cores, have high memory bandwidth and are being widely used as specialized hardware accelerators. A GPU-based parallel LiDAR processing algorithm is implemented with GPU specific memory architecture optimizations. The GPU implementation in this study significantly reduces the processing time of the LiDAR data as compared to CPU-based implementation.