Q&A with Prof. Sayed: Near misses key to reducing traffic accidents

Thursday, December 17, 2015 | By Lindsay Cashin

The numbers associated with traffic accidents worldwide are staggering: 1.3 million fatalities every year, and up to 50 million injuries. They’re also the leading cause of death among 15- to 29-year-olds. But thanks to the work of UBC Civil Engineering professor Tarek Sayed, those numbers could go down.

Sayed, who received the Prince Michael International Road Safety Award Dec. 8, uses automated video analysis techniques to study near misses and propose improvements to intersections and roads to prevent collisions.

How big an issue is road safety?

There are about 1.3 million people killed every year in road collisions worldwide, and for each one of these fatalities there are about 10 injuries. Engineers have always assumed that it’s not our problem; we feel that we are following standards in designing good roads, and any accidents or fatalities are the road user’s fault. Now we know that we can make changes to the system that can prevent driver error.

It’s more cost-effective to address driver error through road improvements than to have a fatality. Can you imagine a mechanical engineer designing a tool that cuts people’s fingers, and saying it’s not a problem with the machine, but with how people are using it?

How do you track and address accidents and collisions?

Our work has been focusing on proactive road safety management, where we make changes to the road to reduce the consequences of a collision and prevent collisions from occurring in the first place. For many years this was done by examining data on traffic collisions, which is a reactive approach. We had to wait two or three years to observe enough collisions before we could analyze and address them. Now, instead of relying on collision data, we rely on data on near misses.

Over the last 10 years we have developed a technique at UBC that does automated road-safety analysis using computer vision techniques at different locations, intersections or highway segments. We see how people move in space and time, and we calculate near misses. It’s very efficient and we get very detailed information that allows us to gain a better understanding of how these near misses and potential collisions occur. Instead of waiting for years of collision data, we can address potential issues quickly and prevent accidents before they occur.