Predicting hazard perception expertise through machine learning

Young novice drivers continue to be overrepresented in crash statistics around the world (OECD, 2019). Hazard perception tests are a key tool used to distinguish between novice and experienced drivers and assessing whether drivers have the necessary competencies to drive safely. The present research aimed to develop a proof of concept to determine the utility of machine learning algorithms in distinguishing between novice and experienced drivers based on eye movement data. Groups of older, experienced drivers as well as younger, novice drivers were asked to watch a 10- minute video clip and their eye movement behaviours were recorded. Results revealed that machine learning algorithms could successfully distinguish between novice and experienced drivers with an accuracy of 71%, with further analysis suggesting that significantly greater accuracies can be achieved. The results hold promise as a potential new Hazard Perception Test (HPT) methodology to objectively assess whether drivers have the necessary visual scanning skills of experienced, more safe drivers, to detect emerging hazards.

• Record URL:
  • https://doi.org/10.33492/ARSC-2023
• Record URL:
  • https://australasianroadsafetyconference.com.au/wp-content/uploads/2023/12/2023_ARSC_Full-Proceedings.pdf
• Supplemental Notes:
  • Extended abstract (researcher)
• Authors:
  • Prabhakharan, P
  • Sivanesan, N
• Conference:
  • Australasian Road Safety Conference, 2023, Cairns, Queensland, Australia
• Publication Date: 2023-9

Media Info

• Pagination: 397-398

Subject/Index Terms

• TRT Terms: Drivers; Eye movements; Forecasting; Hazards; Machine learning; Methodology; Novice drivers; Perception; Vision; Young adults
• Geographic Terms: New South Wales
• ATRI Terms: Eye movement; Forecast; Hazard perception; Methodology; Visual performance; Young driver
• ITRD Terms: 132: Forecast; 9102: Method
• Subject Areas: I83: Accidents and the Human Factor;

Filing Info

• Accession Number: 01931009
• Record Type: Publication
• Source Agency: ARRB Group Limited
• Files: ITRD, ATRI
• Created Date: Sep 17 2024 2:48PM