Exploring risky events and driver responses in automated driving: Insights from real-world videos

• Analysis of risky events which are under-explored and non-fatal precursors to severe incidents • Two-thirds of drivers failed to recognize risky situations before risky events occurred • Crowdsourcing revealed drivers misjudge risk by focusing on severity over probability Automated driving in level 2 autonomy has been adopted by multiple companies such as Tesla and BMW, alleviating the burden on drivers while unveiling new complexities. This article focused on the under-explored risky events-non-fatal but abnormal events like abrupt deceleration, which are frequent precursors to more severe incidents. Through machine learning analysis of 277 real-world videos, we identified that rural roads and complex intersections are primary triggers for risky events, often linked to computer vision errors. A subsequent crowdsourcing survey revealed that two-thirds of drivers fail to anticipate these events, and their perception are influenced by system reactions. This article highlights critical gaps in human-automation collaboration and inform the design of effective automated driving systems.