Cazes Robin, Camps Valérie, Lemercier Céline
Cognition, Languages, Language and Ergonomics (CLLE) Laboratory, University of Toulouse - Jean Jaurés, Toulouse, France.
Toulouse Computer Science Research Institute (IRIT), Paul Sabatier University, Toulouse, France.
Accid Anal Prev. 2025 Jul;217:108051. doi: 10.1016/j.aap.2025.108051. Epub 2025 Apr 18.
Motor vehicle accidents, often caused by human error, remain a significant concern. While automated vehicles have the potential to reduce these accidents by handling driving tasks, unnecessary human takeovers, especially when the automated system is operational, can reintroduce error. This study investigates how situational factors known to trigger anger influence takeover behavior and emotional responses in levels 4 and 5 automated vehicles. Using a driving simulator, 60 participants were randomly assigned to either a goal-aligned condition (clear weather, on-time departure, no traffic) or a goal-conflicting condition (dense fog, delayed departure, slow vehicles). Participants could freely choose between manual and automated driving modes. Results showed a significant increase in takeover frequency, higher negative affect and anger in the goal-conflicting condition compared to the goal-aligned condition. Qualitative data gathered from open-ended questions revealed increased stress and frustration leading to more frequent manual takeovers in goal-conflicting conditions, while participants felt calmer with fewer takeovers in goal-aligned conditions. No link was found between takeover behavior and trust in driving automation. These findings highlight the importance of designing Automated Driving Systems (ADS) that minimize stressors and consider drivers' emotional states to enhance safety and comfort. In this regard, incorporating real-time emotional monitoring and employing cognitive behavioral therapy (CBT) strategies (e.g., situation reappraisal) may help mitigate driver emotional states and prevent unnecessary takeovers.
机动车事故，通常由人为失误导致，仍然是一个重大问题。虽然自动驾驶车辆有潜力通过处理驾驶任务来减少这些事故，但不必要的人为接管，尤其是在自动化系统运行时，可能会再次引入错误。本研究调查了已知会引发愤怒的情境因素如何影响4级和5级自动驾驶车辆中的接管行为和情绪反应。使用驾驶模拟器，60名参与者被随机分配到目标一致条件（天气晴朗、准时出发、无交通拥堵）或目标冲突条件（浓雾、出发延迟、车辆缓慢）。参与者可以在手动和自动驾驶模式之间自由选择。结果显示，与目标一致条件相比，目标冲突条件下的接管频率显著增加，负面情绪和愤怒程度更高。从开放式问题收集的定性数据显示，在目标冲突条件下，压力和挫折感增加导致更频繁的手动接管，而在目标一致条件下，参与者接管次数较少，感觉更平静。未发现接管行为与对驾驶自动化的信任之间存在关联。这些发现凸显了设计自动驾驶系统（ADS）的重要性，该系统应尽量减少压力源并考虑驾驶员的情绪状态，以提高安全性和舒适性。在这方面，纳入实时情绪监测并采用认知行为疗法（CBT）策略（例如情境重新评估）可能有助于减轻驾驶员的情绪状态并防止不必要的接管。
