Modeling driver reaction time in car-following scenarios using physiological data: shape dynamic time warping and generalized linear mixed model approach

Driver reaction time is a key factor affecting traffic flow stability, safety, and adaptive cruise control (ACC) performance, as it directly influences how quickly a driver can respond to potential hazards. This study aims to investigate the influence of leader vehicle characteristics, driving context, and driver stress levels on the reaction time in car-following scenarios. Real-world data were collected using an instrumented vehicle equipped with cameras, GPS, and a wrist-worn physiological sensor. Based on the leader’s relative speed and ego vehicle’s acceleration profiles, reaction time was computed using the shape dynamic time warping (shapeDTW) algorithm. The estimated reaction times were found to follow a lognormal distribution with a mean value ranging from 1.43 s to 1.74 s across the drivers. Results further revealed that drivers reacted quickly under moderate stress levels, as compared to low and high stress conditions. To quantify the effects of traffic and driver-related factors on reaction time, a generalized linear mixed model (GLMM) was developed, results of which indicated that higher ego-vehicle speed reduced reaction time, whereas larger separations and relative speeds of the leaders resulted in longer reaction times. Additionally, low and high stress levels had a significant positive influence on reaction time. The findings provide valuable insights into human response variability and highlight the importance of incorporating drivers’ mental state into ACC design to enhance adaptability and safety in semi-autonomous systems.