Calgary uses AI to determine the 50 highest-risk intersections for pedestrians

The City of Calgary has used artificial intelligence to determine the 50 riskiest intersections for pedestrians across the city, but it’s unclear what will be done to address the issue as just half are recommended for upgrades.

According to a report set to be presented at the city’s Community Development Committee meeting Tuesday, administration used the city’s AI platform to assess and assign a risk score to more than 17,000 intersections across Calgary.

Many of the locations don’t meet “conventional warrant” thresholds for upgrades like traffic signals, overhead flashers, or curb-bump outs, the report said, as that set of criteria is based on factors such as traffic volumes, pedestrian volumes, roadway geometry, and speed limits.

City administration compiled a list of the top 50 highest-risk pedestrian intersections identified by the AI platform that do not have existing traffic signals like rectangular rapid flashing beacons with risk scores ranging from 93.55 to 99.02.

The intersection of Martindale Boulevard and Martindale Gate N.E. topped the list, followed by Falshire Drive and Falton Drive N.E. with a score of 98.54.

Ward 5 Coun. Raj Dhaliwal told Global News he was surprised with the intersections that scored the highest risk ratings, with four of the intersections in the top 10 in his area.

“I was asking myself is there something missing from my assessment,” Dhaliwal said of the intersection in Martindale. “Every time I drive by there, I see less pedestrian traffic.”

Just 26 of the locations are recommended for pedestrian safety improvements which are estimated to cost approximately $6 million, the report said.

According to the report, administration could “proactively advance improvements” with additional funding rather than “waiting” for future traffic counts, collision history or studies to “justify intervention under traditional processes.”

Some on council have questions about how AI was used to determine those findings, including modelling, inputs and prompts.

“I’m all for data-driven decision making. This is empirical evidence that is telling me that these things need to be upgraded,” Dhaliwal said.

The report notes there “may be public scrutiny” around the city’s use of AI, but adds all the recommended improvements were reviewed and validated through professional engineering judgment and technical assessment.

The city’s use of AI is “innovative,” said Alex Gaio, an associate principal at Leading Mobility, with the sheer amount of data administration would need to comb through.

“We have a lot of data, the city updates its hotspot map every 10 minutes,” he said. “So the volume that we get is overwhelming for people to deal with, and we’re racing against risk.”