计算机科学
雷达
人工智能
雷达系统
情绪识别
计算机视觉
实时计算
语音识别
情绪分类
雷达跟踪器
钥匙(锁)
心理学
遥感
人机交互
作者
Chuanliang Shen,Xiaoyuan Ma,Yi Li,Zhenhai Gao,Hongyu Hu
标识
DOI:10.1109/taffc.2026.3739794
摘要
Emotion has been shown to be a key factor influencing drivers' takeover performance (TOP). However, how to effectively utilize driver emotion information to improve TOP prediction remains to be further explored. To this end, this study used millimeter-wave radar to collect drivers' heart rate data under different emotions for TOP prediction. While ensuring privacy protection and physical comfort, multidimensional heart rate features that can reflect drivers' emotions were extracted. Subsequently, a two-stage feature selection method based on affective computing was proposed to identify key heart rate features with strong emotion representation ability, thereby more deeply incorporating driver emotion information into TOP prediction. Specifically, features showing significant differences across emotions were first selected to construct a high-dimensional feature space. Based on this feature space, an emotion recognition model was developed, and SHapley Additive exPlanations (SHAP) was used to evaluate feature importance and further construct a low-dimensional key feature space. Then, using the entropy weight method, we comprehensively evaluated TOP in terms of safety, smoothness, and timeliness, and constructed a TOP prediction model. The results showed that the proposed method can effectively utilize driver emotion information and improve TOP prediction performance. In addition, we further used SHAP analysis to reveal the effects of emotion-related key heart rate features and emotions on TOP. This study provides important insights for the real-time assessment of drivers' takeover capability and offers useful references for designing takeover strategies based on driver emotion awareness.
科研通智能强力驱动
Strongly Powered by AbleSci AI