More than 1200 Australians have lost their lives in fatal vehicle accidents in the past year. Algorithms that predict where and when crashes are likely to happen could help reduce this number.
According to Professor Simon Washington, Head of the University of Queensland (UQ)’s Civil Engineering School, he and fellow researchers were combining video analysis with AI, deep learning and advanced econometrics to measure and detect interactions between road users.
From this analysis, they were aiming to extract data about the times and places where the risk of future accidents was greatest, and determine how to improve roads to prevent accidents from happening.
As well as saving lives, Washington said the technology had the potential to cut the economic impact of road accidents.
“We estimate this technology has the potential to eliminate approximately 540 crashes in South East Queensland each year, equating to about $40 million in reduced crash-related costs,” he said.
Local trial after international success
Along with Associate Professor Md. Mazharul Haque from Queensland University of Technology (QUT) and researchers from the University of British Columbia in Canada, Washington has created algorithms that relate crash risks to the interactions between motor vehicles, pedestrians and cyclists.
Washington said this technology has already been implemented internationally in 20 cities across eight nations, and has led to safety improvements at a number of intersections. He is now working with Queensland’s Department of Transport and main roads to roll out a local trial.
“This pilot project with Transport and Main Roads will help us to build on the knowledge gained from implementation of the technology in other cities,” he added.
According to Simon Harrison, Director (Safer Roads) of Transport and Main Roads, the trial – if successful – could take the department one step closer to putting more engineering road safety solutions into place.
Harrison said that in the past, traffic blackspots were pinpointed through reported crash history, but the road safety sector has shifted its focus to more proactive techniques.
“This research into video analytics has the potential to improve the way we identify safety issues before a crash trend develops,” he added.
Sirens when needed
Startup Waycare Technologies has also released a predictive technology that is being trialled to proactively prevent crashes in the US. Their system combines live data from traffic cameras, sensors and a traffic reporting app with historical data to analyse trends and predict the time and location of future accidents.
This technology has been trialled at intersections in several US cities. When a pattern is identified, police cars are sent to intersections where crashes are likely to occur at the time when they are most probable to happen. The idea is that people will reduce risky behaviour such as speeding if the police are in clear sight.
And it has seen some success. The US media has reported that the technology has led to a 17 per cent reduction in the number of crashes at an intersection in Nevada. According to Waycare, data from their system can also be used to control traffic and avoid congestion caused by car accidents.
Pilot trials of this technology started in 2017 and are continuing, with plans to expand to Europe and Israel next year. The startup has also received a recent $7.3 million funding boost from venture capitalists.