Advancing LiDAR perception in degraded visual environments: a probabilistic approach for degradation analysis and in inference of visibility

Overview & Synopsis

The subject focuses on the adverse weather limitations of autonomous vehicles. Weather events like rain, fog or snow alter the performances of our current sensors by generating false positives and degrading signals. Thus, the vital capacities of autonomous vehicles of obstacle detection, localization and navigation are reduced. To be brought to the market with full disponibility, these issues must be overcome. A study of the physics of the sensors against these phenomena will allow us to understand the limitations. New available sensors and signal processing methods will lead us to an innovative sensor set coupled with adapted algorithmic solutions of filtering and data fusion. The goal is to enhance the perception capabilities in adverse weather conditions to improve autonomy availability.