How AI is making Australia's roads safer | Jan20 Newsletter
Technology has always been at the forefront of reducing car crashes – think red-light cameras, speed detectors or the advanced scanning equipment inside a highway patrol car. Now the Western Australian (WA) Government has added artificial intelligence (AI) in the fight against road crashes – and machine learning is selecting sites for red-light cameras.
How AI is improving safety across the globe
In the 12 months to October 2019, just over 1,200 Australians lost their lives on the road. So it’s no wonder that Australia, like many other countries across the globe, are trialling and implementing a range of AI technologies to help identify road crash hotspots before they happen.
The trials have been encouraging so far. In one stretch of Nevada interstate highway, authorities give AI the credit for a 17 percent reduction in crashes. In Queensland, one AI trial is hoping to reduce the number of crashes by 540 annually.
It’s all about anticipating crashes before they happen.
The WA program
When the Western Australian Road Safety Commission (RSC) set the ambitious target of drastically reducing the road toll in its Towards Zero 2008–2020 program, they set their sights on one troubling statistic: that 50 percent of road crashes happen at intersections.
50 percent of road crashes happen at intersections.
But looking back at the road data on killed and seriously injured (KSI) at intersections was both time-consuming and labour-intensive. It also had one fundamental flaw: it was looking backwards. By contrast, the RSC wanted to explore how they could prevent accidents in the future.
Enter: machine learning.
Machine learning and red-light cameras
The WA Government used SAS machine learning to calculate the probability of road crashes at intersections in the future and the government could respond by placing red-light cameras in potential hotspots. AI provided several benefits.
First, it could calculate the data at a rate of five times faster than humans and, more importantly, it was looking forwards not backwards. The machine could assess the risk for each intersection in the future, rather than counting the number of crashes in the past.
The machine could assess the risk for each intersection in the future, rather than counting the number of crashes in the past.
Once the data had been crunched, red-light cameras were placed at the intersections most likely to be crash sites – a clear signal for motorists to slow down.
It’s estimated that the technology will reduce road crashes at intersections by 25 percent, and the pilot has been so successful that the WA Government has brought the analytics in-house rather than hiring outside consultants.
Clearly, AI will play a leading role in the fight to make our roads safer. It’s just one of the many ways AI is helping governments serve us all better.