고신뢰 자율주행을 위한 악의 조건 측위성능 평가 시나리오 설계 방법 연구

Abstract

In order for autonomous vehicles to drive safely on real roads, the role of high-definition maps is becoming increasingly important, and so is the importance of positioning technology. Since precise localization is an indispensable element of autonomous driving systems that recognizes the vehicle’s surroundings and determines its current location, the performance of localization should be evaluated not only in normal driving situations, but also in harsh driving situations that limit the ability to perceive the surroundings. In this study, we propose a methodology for building risk scenarios for the evaluation of localization performance in adverse driving conditions. To accomplish this, we define adverse conditions that affect the positioning performance of autonomous systems, break them down into perception related causal factors, and specify the effect of each on the physical and functional properties of sensors to derive risk scenarios caused by adverse conditions. In addition, by combining various scenarios with sensor specific perception limitations to build an integrated scenario, we found that realistic and efficient performance evaluation is possible. © 2023 KSAE.