Evaluating the Feasibility of Integrating Webcam-Based Eye Tracking into Hazard Perception Assessments

Hazard perception is fundamental to driver safety, yet current assessment methods, including those employed by the UK Driver and Vehicle Standards Agency (DVSA), rely primarily on reaction-time-based scoring and ofer limited insight into the attentional processes underlying hazard detection. Although eye-tracking research has demonstrated the importance of visual attention in hazard perception, traditional approaches depend on specialised hardware, restricting their scalability and integration into real-world assessment environments. This study presents a proof-of-concept investigation into the integration of webcam-based gaze estimation within a hazard perception testing framework. A prototype system was developed using MediaPipe to estimate gaze direction during video-based hazard perception tasks, enabling the capture of low-resolution attentional signals using standard consumer hardware. Ten domain-informed participants, all software engineers within the UK Department for Transport, completed the task and provided structured feedback through a mixed-methods evaluation. Findings demonstrate strong support for the integration of gaze-based feedback, with all participants indicating that such information would enhance hazard perception training or testing. Despite recognised limitations in accuracy, calibration stability, and sensitivity to environmental conditions, participants considered coarse gaze data sufficient to provide meaningful insights into attentional behaviour. The results highlight a clear distinction between the requirements of high-stakes assessment and training applications, with gaze estimation showing particular promise as a diagnostic and feedback tool. This research contributes an initial validation of the feasibility of incorporating low-resolution gaze estimation into hazard perception systems in a manner that is interpretable, accessible, and aligned with existing assessment constraints. The findings establish a foundation for future work exploring gaze-informed feedback, improved calibration techniques, and the development of scalable, attention-aware driver training systems.

Name dc:type.qualificationname

msc

Level dc:type.qualificationlevel

masters

Grantor dc:publisher.institution

University of Wales Trinity Saint David

Year dc:date.issued

2026

Author dc:creator

• Evans, Ryan

• QA76 Meddalwedd Cyfrifiadurol

Dc Identifier Grantnumber

UWTSD

OAI identifier oai:identifier

oai:repository.uwtsd.ac.uk:4314