"Driving data of autonomous vehicle"

"This paper proposes a real-world data-driven framework for evaluating the automated vehicles (AVs) driving safety on expressways. To develop the framework, 31 evaluation items and 35 evaluation metrics were derived from the Road Traffic Act, UN Regulation No. 157, and the Safety Standard for Partially Automated Driving Systems (i.e., Safety Standard in Korea). A six-step evaluation process was designed to enable real-world safety evaluation. Real-world driving data from an AV operating on an expressway were analyzed to validate the proposed evaluation framework. Based on the automated driving system (ADS) functions, 18 evaluation items out of the 31 total items were selected for validation. Based on the results, cases where the ego vehicle did not partially meet the safety criteria for \u201cSpeed limit compliance evaluation\u201d and \u201cSafe distance compliance evaluation\u201d were identified. Further analysis revealed that the speed-limit violations were brief or occurred during transitions to higher-speed-limit sections, suggesting minimal impact on safety, whereas the safe-distance violations were caused by abrupt lane changes by surrounding vehicles, with the ego vehicle\u2019s responses varying according to the relative speed. These findings confirm that the proposed framework effectively evaluates AV driving safety under real-world conditions and provides insights into driving behavior."