Go to the Globe and Mail homepage

Jump to main navigationJump to main content

Two firefighters by a car wreck. (Lane Erickson/Getty Images/iStockphoto)
Two firefighters by a car wreck. (Lane Erickson/Getty Images/iStockphoto)

Forward Thinking

Data crunching can prevent cars from crashing Add to ...

Having already cut traffic collisions resulting in injuries and deaths nearly 40 per cent in five years by analyzing patterns from data it has collected, the city of Edmonton is using predictive technologies to increase road safety even more.

As drivers navigate Edmonton’s streets, they will inevitably pass a digital messaging sign, one of as many as 200 strategically located throughout Alberta’s second-largest city.

More Related to this Story

“On a day when there would be a high risk for collisions based on road conditions – really slippery or very cold temperatures – you would see a message advising, ‘Road conditions icy ahead,’ or that type of thing,” says Gerry Shimko, executive director of the City of Edmonton’s Office of Traffic Safety (OTS).

The digital signs are just one element of an innovative traffic safety program that has dramatically reduced vehicle collisions in the Edmonton region since OTS launched in late 2006. Underpinning this effort is data analytics. Having used analytics to do everything from identify dangerous road design to predict poor driving conditions, OTS plans to integrate multiple data sources so it can create a comprehensive real-time picture of city traffic.

Mr. Shimko says OTS has two main focuses: speed management; and research into road user behaviour, urban traffic safety and data analytics.

OTS’s staff of 27 includes a seven-member data analytics team. “One of our fundamental principles in the office is an evidence-based approach to traffic safety,” Mr. Shimko says. “You need to have statistically significant analytics to augment that kind of approach.”

For Edmonton, whose metropolitan population is roughly 1.2 million, the results are impressive. In 2007, the city had 28,521 motor vehicle collisions, of which 5,513 involved injuries and deaths, according to the OTS. Last year, those totals fell to 23,168 and 3,385, respectively – a 38.6-per-cent reduction for the latter. This decrease is all the more remarkable given that, between 2006 and 2011, Edmonton’s population grew more than 11 per cent, according to Statistics Canada.

Starting in 2007, one of the first things OTS did with data analytics was pinpoint parts of the city’s infrastructure where there were more collisions, Mr. Shimko says. It found that 30 per cent of collisions took place at major intersections, which had dedicated lanes for making right turns. These types of turns could cut down on visibility for drivers.

So Edmonton moved from one-size-fits-all road design to a flexible approach that lets OTS consider factors such as what kinds of vehicles and how many pedestrians use a particular intersection.

The city has been replacing those perilous right-turn lanes with different types that improve visibility, from simple turn lanes with shallower angles to dedicated ones, depending on what kinds of vehicles use them.

Beyond looking for existing patterns to improve traffic safety, OTS has started peering into the future, with the use of predictive analytics.

Last year, its analytics team collaborated with the University of Alberta’s departments of Civil and Environmental Engineering, and Earth and Atmospheric Sciences, to build a computer model that shows, based on snow conditions, when to expect a higher number of collisions.

This weather model predicts collision trends seven days in advance with about 90-per-cent accuracy, Mr. Shimko says. As conditions turn hazardous, the city’s Traffic Management Centre can use this information as part of its digital sign messaging.

As well, OTS works with the Edmonton Police Service on several initiatives to reduce speeding and other traffic violations.

Last year it began collecting and analyzing data on every licence plate detected via intersection safety devices, photo speed enforcement and other automated means, and giving high-risk drivers’ plates to the police. “Then the police can target their resources based on those high-risk drivers,” who typically have more injury collisions, Mr. Shimko says. “Some of the initial results are quite promising.”

Edmonton is also collaborating with the University of Alberta’s Centre for Smart Transportation on a project to improve traffic flow on Whitemud Drive, a major east-west artery whose 100,000 daily vehicles make it the city’s busiest traffic corridor. It has installed sensors under the pavement that relay data such as speed and vehicle occupancy to a central system at the Traffic Management Centre, says senior traffic engineer Ken Karunaratne.

“The software analyzes the data, and then, if there’s any congestion based on vehicle spacing or any other key indicators, it will suggest an alternate speed so that we can still maintain vehicle movement at the lower speed,” Mr. Karunaratne says. When the project is up and running, its operator will be able to activate variable speed signs along Whitemud Drive.

Zhi-Jun (Tony) Qiu, a University of Alberta assistant professor of engineering who directs the Centre for Smart Transportation, has worked with OTS since 2009. Besides participating in the Whitemud program, Dr. Qiu is studying Edmonton’s aggregated cellphone use data to explore the relationship between collisions and traffic flow, volume and speed. “This kind of information will have a huge influence on potential incidents,” he says.

Looking ahead, OTS wants to establish a transportation intelligence centre, Mr. Shimko says. His office is looking at buying a platform that would let it bring together information from road sensors, closed-circuit TV cameras and other data sources to provide what he calls real-time situational awareness.

Mr. Shimko gives the example of Edmonton’s bus service, which has 850 vehicles on routes at any time of day. If OTS gathered data on buses’ slippage, it could develop an algorithm to identify difficult snow conditions. Meanwhile, collecting data from road maintenance crews would reveal how effective their efforts were. “Just by integrating those two very simple data streams, we can them help identify…which roads do we need to be on, where should we get rid of the snow first, where are those collision problems,” Mr. Shimko says.

Sharing data across departments can have powerful effects.

“We know from working with the police in a really close relationship that we could see a dramatic drop in collisions just from giving them information on where to do enforcement,” Mr. Shimko says. “We have some really, really good results when those synergies happen.”

 

Follow us on Twitter: @GlobeBusiness

In the know

Most popular video »

Highlights

More from The Globe and Mail

Most Popular Stories