A method for the estimate erroneous fog detection in automotive LiDAR

LiDARs are crucial for fully autonomous driving and have greatly improved in performance and availability on the market, leading to a need for methods and tools to compare them. Most of the studies in this field consider only tests in optimal meteorological conditions. Otherwise, it is relevant to investigate the performance of these devices in adverse weather conditions. In this study, we extend our work regarding the characterization of LiDARs in a foggy environment. In particular, we introduce a new measurement method aimed at estimating the minimum fog concentration for which a LiDAR detects the fog as a target, precluding the possibility of detecting any other objects that may be present inside the bank of fog. This method, together with the previously presented method, has been used to characterize and compare two commercial LiDARs - the MRS 6000 by Sick and the VLP 16 by Velodyne. These analyses were achieved through a custom setup extremely simple to be implemented. Thanks to the proposed methods, it has been possible to discover significant differences in the performance of the two LiDARs in foggy environments. For example, the VLP 16 showed good performance starting from optical visibility of about 3 m, while, for the MRS 6000, a visibility of about 40 m is required.