WS21: Driver Vigilance Estimation for Vehicle Active Safety

Workshop Code: 16qc3


  • Bao-Liang Lu
    Affiliation: Shanghai Jiao Tong, China

  • Chin-Teng Lin
    Affiliation: University of Technology Sydney, Australia

Scope and Goals

Vigilance decrement or attention lapse has long been recognized as the critical factor responsible for thousands of deaths and injuries each year in the public traffic community. Driving tasks, particularly truck driving and high-speed trains, require sustained high vigilance. However, efficient techniques for quantifying driver vigilance levels are still lacking, which leads to the inability to provide active feedback for active safety systems. Although considerable progress has been achieved in various areas over the past decades, accurately estimating driver vigilance in real-world driving environments is still difficult. The main reason for this difficulty is that vigilance states are intrinsic mental states that involve temporal evolution rather than a time point. It is difficult to evaluate mental states without using an intrusive stimulus or behavior probe. Moreover, real-world applications require continuous vigilance estimation with high temporal resolution. Vigilance decrement is typically accompanied by both external behaviors, such as head nodding, yawning, and eye closure, and internal physiological changes. Various approaches based on these cues have been developed. Among these various modalities, physiological signals have been found to be relevant for different vigilance levels. However, how to identify reliable and valid biomarkers remains a challenge within the research community. The aim of this workshop is to give a forum for researchers to present the state-of-the art of neural mechanism, modelling, devices, and systems for driver vigilance estimation and to exchange ideas and issues.