Article Text

Download PDFPDF

250 Pedestrian-vehicle interactions: early results from the Australian naturalistic driving study (ands)
  1. Garrett Mattos1,
  2. Raphael Grzebieta1,
  3. Ann Williamson1,
  4. Jake Olivier2,
  5. Jan Eusebio1,
  6. Wu Yi Zheng1,
  7. John Wall3,
  8. Jude Charlton4,
  9. Mike Lenné4,5,
  10. Jack Haley6,
  11. Ben Barnes7,
  12. Andry Rakotonirainy8,
  13. Jeremy Woolley9,
  14. Teresa Senserrick1,
  15. Kristie Young4,
  16. Narelle Haworth7,
  17. Mike Regan1,
  18. Samanatha Cockfield10,
  19. David Healy11,
  20. Antonietta Cavallo11,
  21. Marilyn Di Stefano11,
  22. Hee Loong Wong12,
  23. Iain Cameron13,
  24. Michael Cornish14,
  25. Christine Baird15
  1. 1­Transport and Road Safety (TARS) Research, University of New South Wales, Australia
  2. 2School of Mathematics and Statistics, University of New South Wales, Australia
  3. 3Centre for Road Safety, Transport for NSW,Australia
  4. 4Monash University Accident Research Centre (MUARC), Australia
  5. 5Seeing Machines, Braddon, ACT,Australia
  6. 6National Roads and Motorist Association (NRMA), Australia
  7. 7Centre for Road Safety, Transport for NSW, Australia
  8. 8Centre for Accident Research and Road Safety – Queensland (CARRSQ), Australia
  9. 9Centre for Automotive Safety Research, University of Adelaide, Australia
  10. 10Transport Accident Commission, Victoria, Australia
  11. 11Vicroads, Australia
  12. 12Hyundai Australia
  13. 13Office of Road Safety, Main Roads Western Australia
  14. 14Motor Accidents Commission, Adelaide
  15. 15Motor Accidents Authority, Australia


Background Typologies have been defined previously for pedestrian-vehicle interactions and are primarily based on retrospective analysis of crash data. The naturalistic driving study currently underway in Australia makes it possible to study pedestrian-vehicle interaction events that would not otherwise be identified in the crash data. This work evaluates the feasibility of using automated, manual, and semi-automated methods to identify pedestrian-vehicle interaction events.

Methods Sensors and cameras were installed on the vehicles of volunteers in and around two major Australian cities which recorded their natural driving behaviour for 4 months. Forward video from select vehicles was reviewed independently by two reviewers to identify potential pedestrian-vehicle interaction events from which a typology of behaviours was formulated. These events served as the gold standard against which select automated and semi-automated methods of identification were assessed.

Results A prototype typology of pedestrian-vehicle interaction events was formulated using naturalistic driving data and categorised in terms of risk of being struck. Some case scenarios will be discussed. The feasibility of using select automated, semi-automated, and manual methods to identify these events was also evaluated.

Conclusions This work provides a first look at using Australian naturalistic driving data to study the interactions between vehicles and pedestrians. These findings will assist in the development of methods that can be used to most effectively answer research questions pertaining to interactions between vehicles and pedestrians as well as other vulnerable road users in the future.

  • Pedestrian
  • Crash
  • Naturalistic Driving Study
  • ANDS
  • Typology

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.