Developing a Metric to Optimize LiDAR Scan Parameters: A Controlled Experimental Approach Using Boston Dynamics’ SPOT

Light detection and ranging (LiDAR) sensor systems provide valuable geospatial data without complete reliance on visual or tactile data. However, optimizing data collection for these systems is challenging due to the lack of standards for distinguishing scan quality. This research defines a performance metric for a high-quality scan to standardize LiDAR setup across diverse applications. To test this metric, a 3D LiDAR sensor was mounted on Boston Dynamics’ SPOT to collect data in a controlled environment, altering the scanning parameters to generate a broad set of 3D point clouds. The scans were evaluated with two statistical comparison metrics to determine the optimal configuration and the effect of each parameter on the final scan quality. The performance of a 3D object classification neural network was closely correlated to the performance of our metric, validating its applicability in real-world use cases.