Real-Time Autonomous Vehicle Sensor Performance Assessment in Adverse Weather Conditions
Stanislav Svediroh, Ludek Zalud
Abstract: The future of the automotive industry appears to be intricately linked to Advanced Driver Assistance Systems (ADAS) and various levels of Automated Driving Systems (ADS). Over the years, numerous companies have incorporated sensors into their vehicles, however, none have yet achieved the development of a completely robust and self-aware system capable of operating safely in adverse weather conditions. To guarantee safety, the vehicle must possess an awareness of its environment and the current performance of its sensors. This includes the ability to detect not only current weather conditions such as rain, fog, haze, and snow, but also smoke, soiling from various sources, and extreme lighting conditions such as glare or low light. It is crucial for the vehicle to detect these conditions in real-time without delaying decision-making systems. This study summarises the effects of various environmental threats on commonly used sensors in ADAS or ADS and proposes algorithms to detect degrading sensor performance, which can then be integrated into the sensor fusion framework utilised in the creation of the vehicle’s local map. The ultimate aim of such a system is to accurately detect and report sensor degradation, enabling subsequent sensor fusion and path-planning algorithms to modify the vehicle’s behaviour and minimise unreasonable risk. Index Terms—ADAS, ADS, Adverse Weather, Sensor Performance Assessment
Real-Time Autonomous Vehicle Sensor Performance Assessment in Adverse Weather Conditions
