Cruise Control: AI Steering Public Health in Road Safety

Onions BC, Sharma S, Gerber Z, Mendz GL, Suttie MC, Suttie JJ

Abstract

Data from the Australian Bureau of Infrastructure and Transport Research Economics paints a stark picture with over 1,000 recorded road fatalities being annually recorded in recent years (Bureau of Infrastructure and Transport Research Economic, 2021). Motor vehicle crashes in rural Australia are a significant cause of morbidity and mortality, with fatal crashes being almost five times more likely in rural than in urban areas². Furthermore, the types of crashes and speed of travel tend to be more dangerous in these areas and thus contributing to a higher fatality rate. As such, road safety is a key research priority for rural medical schools. Key contributors to this higher rate of road traffic accidents in regional areas include the higher speeds, poorer road surface, fatigue, longer journey time and limited enforcement (Department of Infrastructure, Transport, Regional Development, Communication and the Arts, 2021). 

 Autonomous vehicles offer some promise in addressing the issue of road safety in regional areas. These vehicles range from partial automation, for example with lane-assist as is already available in some vehicles in Australia, to full automation where all aspects of driving are performed without need for human input. For example, with 30% of patients who were hospitalised after a motor vehicle accident reporting being distracted prior to the crash (Sheehan et al., 2008), autonomous vehicles offer great promise, not being prone to distraction as their human counterparts are. Human response times to traffic hazards are much slower than those of computer-based technologies with respective response times of 1.5 to 2.5 seconds (Austroads, 2021) compared to 0.05 to 0.36 seconds (Wang et al., 2020). 

Despite this great promise of autonomous driving features, studies predict that in a challenging driving environment and inclement weather, human drivers may prove to be more adept at driving safely than autonomous vehicles (Karla & Paddock, 2016). This is significant as 22% of all semi-automated vehicle crashes have been identified as occurring in wet road surface conditions as shown in data released by The United States Department of Transportation. Whilst autonomous vehicles may play a significant role in enhancing road safety, the ability of these vehicles to effectively function on Australian roads and environment is still unknown. Autonomous vehicle trials are currently underway in Mount Isa in regional Queensland to test the suitability of road conditions and infrastructure for autonomous vehicles. 

While crashes arising from human error could be reduced by autonomous vehicles, there are numerous potential risks associated with this emergent technology. Our review of several datasets and international accident registries suggests risk factors like weather and poorly sealed and mapped roads, particularly in rural and regional Australia, could play a disproportionately significant role in conferring risk in autonomous vehicle technology and should inform Australian policy in this area.

References

1. Austroads (2021). Guide to Road Design, Guide to Road Design Part 3: Geometric Design. Sydney, Australia: Austroads Ltd.
2. Bureau of Infrastructure and Transport Research Economics (2021). Road trauma Australia 2021 statistical summary.https://www.bitre.gov.au/sites/default/files/documents/road_trauma_2021.pdf.
3. Department of Infrastructure, Transport, Regional Development, Communications and the Arts.(2021). Factsheet: Regional road safety [Fact Sheet]. National Road Safety Strategy. https://www.roadsafety.gov.au/nrss/fact-sheets/regional-road-safety
4. Kalra, N., & Paddock, S. (2016). Driving to safety: How many miles of driving would it take to demonstrate autonomous vehicle reliability?Transportation Research Part A: Policy and Practice, 94, 182–193. https://doi.org/10.7249/rr1478
5. National Highway Traffic Safety Administration. (n.d.). Standing General Order on Crash Reporting. https://www.nhtsa.gov/laws-regulations/standing-general-order-crash-reporting
6. Sheehan, M., Siskind, V., Turner, R., Veitch, C., O’Connor, T., Steinhardt, D., Blackman, R., Edmonston, C., & Sticher, G. (2008). Rural and Remote Road Safety Study Final Report. Centre for Accident Research & Road Safety - Queensland (CARRS-Q).
7. Wang, L., Fan, X., Chen, J., Cheng, J., Tan, J., & Ma, X. (2020). 3D object detection based on sparse convolution neural network and feature fusion for autonomous driving in Smart Cities. Sustainable Cities and Society, 54, 102002. https://doi.org/10.1016/j. scs.2019.102002