Safety-First Autonomous Vehicle Technology: Empirical Assessment of Sensor Performance in Diverse Environmental Conditions

Many companies and institutions focus on autonomous vehicles. Accordingly, the commercialization of fully autonomous vehicles is expected to proceed rapidly. Autonomous vehicle companies are already demonstrating or commercializing vehicles on real roads. However, autonomous vehicle data is not disclosed to the public, and related academic research is very rare. In this study, changes in the functional levels of autonomous driving sensors in various environments were analyzed using actual collected autonomous driving data. A hypothesis was established by reviewing previous studies related to the functional levels of sensors, and the hypothesis was verified by comparing actual autonomous driving data using two different statistical analysis methods. The hypothesis was tested using the 2-sample K-S test and the DTW algorithm, with different sensor elements used in each verification stage. As a result of the analysis, we found that sensor performance changed on rainy and cloudy days compared to sunny days. This study confirms that the functional levels of autonomous vehicle sensors change depending on the environment. The results of this study are expected to serve as foundational data for establishing standards and criteria for safety evaluations of autonomous vehicles in the future.