Answers are grounded in this paper and cite their sources.
This study investigated the factors influencing two common hazardous driving events (HDEs) in AVs, sudden acceleration and sudden deceleration, using real-world data collected from eight Level 3 AV taxis operating at the Sangam testbed in Seoul, Korea, covering 2.96 million seconds of autonomous driving. An analytical framework combining cluster correspondence analysis (CCA) and multilevel multinomial logistic regression (ML-MLR) was applied. The CCA identified four distinct behavioral clusters, revealing that HDEs are not randomly distributed but instead associated with specific operational contexts. The ML-MLR results quantify the effects of key factors, indicating that land use types, lane configuration, and roadside infrastructure significantly influence HDE occurrence. Notably, three-lane roadways provide the most stable operating conditions, while bicycle lanes and transit stops increase longitudinal speed variability. These findings highlight condition-dependent nature of AV behavior and provide important implications for infrastructure design and adaptive control strategies to enhance safety and operational efficiency of AVs.
No takes yet. Share an insight, caveat, or question.
Nguyen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: