Many drivers have stories about near-miss collisions or situations where a crash was narrowly avoided -- information that’s invaluable to experts who are looking at it more and more to help make roads safer.

It's a technique being used right here in Manitoba to give traffic engineers a glimpse into the future and intervene before it's too late.

About a year and a half ago, Winnipeg-based road safety engineer Craig Milligan co-founded the firm called MicroTraffic.

Milligan along with Joel Penner and Loren Dueck formed a team of 16 people to help cities, provinces, states and countries, which seek their services, better understand safety at intersections.

"We're looking for situations where a serious injury or fatality was narrowly avoided,” said Milligan. "They're looking for abnormally elevated risk levels so that they can get out and proactively make an improvement before a serious fatality or injury happens."

“Sometimes you need to prove that a problem exists before you can get money to fix it.”

Traditionally, collision data has been used but engineers have long recognized shortfalls of analyzing that information due to small sample sizes.

It's why there's been a move towards improving intersection safety by studying near misses.

Inside MicroTraffic’s operations lab in south Winnipeg, its team of engineers looks at close calls on the road.

A proactive instead of a reactive approach to curbing crashes.

"We can predict future safety of the site before accidents actually occur,” said Penner, MicroTraffic’s manager of engineering operations. "Regularly, we hear about people being seriously injured or killed and I don't think it's a necessary byproduct of the system and I'm really proud of how we're able to use our tools to make life better for everyone in a really recognizable way."

The company uses video from traffic management centres or specially-mounted cameras to study intersection safety.

Engineers load the footage into computer software powered by artificial intelligence, programmed to run through one frame of video at a time to look for unsafe situations.

"So for example one vehicle quickly passes in front of another, that's an indication of maybe some type of unsafe traffic conflict and that's what we extract and use to look at safety,” said Dueck, MicroTraffic’s manager of development and computer vision.

Manitoba Infrastructure said over the past year MicroTraffic has been used to detect near misses on provincial roadways.

Information gathered has led to median closures and access management along the south Perimeter Highway.

"We thought there was a problem there and this was just one more way we could quantify the magnitude of that problem," said Manitoba Infrastructure director of traffic engineering Glenn Cuthbertson.

Studying near misses to predict collisions started decades ago.

A Canadian researcher who's helped develop the approach said advancements in technology have made the technique more feasible to do on a larger scale.

"We can identify near misses and our research has shown that there is strong correlation between near misses and expected number of collisions at any site,” said University of British Columbia civil engineering professor Tarek Sayed. "We can understand why near misses happen very quickly and intervene to eliminate these near misses. The traditional approach is relying on collision police reports and these do not say that much and they are after the fact."

If the number of near misses leads to changes, engineers can then use the technology to further study an intersection to determine if the modifications are making a difference.

Manitoba Infrastructure said the technology was used to study two locations along the south Perimeter and has also been used to look at two other intersections but at this point the province won't say which ones.

The city of Winnipeg isn’t using traffic cameras to detect near misses on roadways but a city spokesperson said there’s interest in exploring the potential to improve safety by leveraging investments made in the Transportation Management Centre which contains 70 cameras.