On Thanksgiving night a few years ago, Ling Zhuang’s husband was driving along 124th Avenue Northeast in Bellevue when he collided with another car at Northeast Eighth Street. It was dark and raining, and he was confused by the traffic-light signals, she said.

In the months that followed, more drivers crashed at and near the intersection.

“That made me realize that it’s probably not just my husband, but quite a few people who had this issue,” she said.

Traffic Lab is a Seattle Times project that digs into the region’s thorny transportation issues, spotlights promising approaches to easing gridlock, and helps readers find the best ways to get around. It is funded with the help of community sponsors Alaska Airlines, Kemper Development Co., Madrona Venture Group, NHL Seattle, PEMCO Mutual Insurance Company and Seattle Children’s hospital. Seattle Times editors and reporters operate independently of our funders and maintain editorial control over Traffic Lab content.

So she wrote to the Bellevue Department of Transportation last summer asking for a light for drivers turning left from 124th Avenue Northeast onto Northeast Eighth Street toward downtown.

Since then, Bellevue — in partnership with the transportation-technology firm Transoft Solutions and the nonprofit organization Together for Safer Roads — has examined the safety of that intersection and 39 others using traffic cameras and artificial intelligence technology.

Advertising

The analysis of 5,000 hours of traffic-camera video and millions of trips by car, bike and on foot found that near misses in which crashes nearly occur are a reliable predictor of where future collisions will happen.

As cities work to make progress on their Vision Zero goals, an effort to eliminate traffic-related deaths and serious injuries by 2030, the results of Bellevue’s project, released last month, represent a proactive approach.

Last year, 25 people were seriously injured and five were killed while driving, biking or walking in Bellevue, the city says.

Data from the video footage captured traffic volume, vehicle and bicycle speed, and crashes and near misses — information the city has used to identify problematic intersections and propose improvements.

“Currently, jurisdictions rely on police-reported events to inform where they make improvements. That’s problematic. Those crashes are rare and random, and it often takes five-plus years for a pattern to emerge. During that time, you have all of these people being injured,” said Franz Loewenherz, principal transportation planner with the city of Bellevue.

“Vision Zero is about imagining a future where people don’t have to get hurt in order to intervene proactively to bring about safety,” he said.

Advertising

What the analysis found

The one-week study period last September found 20,000 so called critical conflict interactions — near misses in which roads users came within two seconds or less of colliding and at least one road user had to act to avoid a crash.

Most of those critical conflicts — 97% — occurred between two drivers. However, bicyclists were 10 times more likely to be involved in a critical conflict than motorists despite representing only 0.1% of road users, the analysis found.

Vicky Clarke, policy director at Cascade Bicycle Club who oversees Eastside transportation, said the data is unsurprising in showing that bicyclists are vulnerable road users at higher risk.

She applauded the data-collection effort but said what’s needed is a commitment to change.

“You can do all the data collection in the world, but what you really need to do is invest in safety improvements in infrastructure. That’s the only thing that’s going to make a difference,” she said.

The analysis also found that more than 10% of drivers were speeding, and half of those drivers were traveling more than 11 mph over the speed limit.

Advertising

Driver speeds were higher in residential areas and locations outside downtown, the report found. While locations with higher posted speed limits saw drivers traveling at higher speeds, posted speed limits had no effect on how much speeding occurred.

Information from the study was not used for enforcement purposes. No one received a ticket during the assessment period, but Loewenherz said it does give the city “the granularity of insight” for when and where speeding occurs and the magnitude of the problem.

Data also showed the most frequent critical conflicts occurred between vehicles turning left and those heading straight through an intersection.

Armed with more detailed information about the intersection at 124th Avenue Northeast and Northeast Eighth Street, Bellevue added a left turn signal so drivers turning left receive a green arrow first, and then can still turn on a flashing yellow arrow when drivers going through the intersection have the green light.

Previously, people making left turns always had to yield and didn’t have a green arrow.

After the city made signal changes last fall, at a cost of $10,000, critical conflict interactions decreased by 60%, city data showed.

Sponsored

How the project works

Bellevue relied on its high-definition traffic cameras to stream video into a central control center.

Of the 40 intersections in the study, 31 were in the city’s High Injury Network — areas where serious and fatal crashes have occurred most often. The others were chosen based on geographic diversity, land uses, population density and road design.

As video streams in, Transoft builds a profile for every location to understand the users most at risk, the interactions occurring and the speeds recorded when an interaction happens. That provides city engineers “concrete evidence to quantify that a proposed recommendation will improve the location,” said Charles Chung, the vice president for transportation safety at Transoft.

The Bellevue Transportation Department prepared a proposal to improve the intersections identified as most problematic in the study.

“We have to wait to see how it fares within the budget process, but the intent there is to do an iterative process where we get some improvements out there rapidly,” Loewenherz said.

“By leveraging this data, we don’t have to wait for five years for a pattern to emerge. We can run a one-week assessment period and make some improvements and then run an after assessment and derive insight that it really resulted in a favorable outcome,” he said.

Advertising

What other cities can learn

Bellevue made an attractive partner because it’s a progressive city that’s been receptive to new technologies and concepts, Chung said. The city had a 10-gigabyte communications network to process the footage, which was captured by cameras already in place.

That meant there was no added capital costs to the city for the project, Loewenherz said. Bellevue’s expense came from staff time to talk through system requirements and co-author the reports. Together for Safer Roads also provided some funding.

While not all cities have a 10-gigabyte network, this same kind of program is feasible with a lower-speed network, Loewenherz said. And if a city doesn’t have traffic cameras in place, it’s possible for cities to have cameras mounted at specific locations for limited duration assessments.

Chung said the team did run into some challenges along the way, like initially underestimating the computing power required to process the video. They eventually acquired a high-performance computer to conduct the work.

Noah Budnick, the senior director of programs and operations at Together for Safer Roads, said he hopes to see this kind of collaboration replicated in other cities to work toward achieving Vision Zero.

“If you’re basing your Vision Zero actions on fatality and serious injury data, then you’ll never achieve your goals. You’re always one step behind,” he said.