Case study: Transurban uses automation to detect road incidents

Global road toll operator Transurban monitors and controls toll networks across Australia, its intelligent transportation system (ITS) helps the company manage some of the busiest roads in the country.

Nick May, head of ITS strategy and innovation at Transurban explains some of the more recent digital improvements that have happened at Transurban.

Transurban uses a system called Automatic Vehicle Incident Detection or AVID, a camera-based surveillance system that detects any stopped vehicle, vehicle on fire or debris on the road.

This system has been used since 1999 beginning with the City Link in Melbourne but May said they are constantly trying to improve its performance.

“We get false alarms, we don't generally miss anything, but we do get quite a lot of noise around false alarms. That can be anything from weather-related where there's light coming into the portal or water on the ground, it can just confuse the system,” he said.

To improve the system, they are focusing on data fusion.

“There's been developments around how do we improve that performance and make sure operators get the best possible information to act on. Rather than just rely on that pure camera stream, we're trying to aggregate other data sources and triangulate around the decision.

“The camera system might say there is a stopped vehicle but actually third party, flowing vehicle data from TomTom and GPS data suggest that everything's travelling perfectly fine. There's no slowdown in traffic. Chances are, we've got a misread there,” he said.

As they add more and more layers of data, they can aggregate around driving up that performance, May explained.

“The natural evolution of that is to look at the video system itself and how can we apply more sophisticated video analytics to that video stream. We’re starting on the journey of that at the moment, essentially looking at, we collect so much video data across all of our cameras and all of our assets, we must be able to do more with it,” he said.

“There is such rich data there and we say if we can see it, we can measure it, and we can do something with that. So that could be again, just improving incident detection. But we're finding more different use cases that we can take that rich data, video data and turn it into a useful either traffic management use case or in some instances a business-facing use cases.”