Our Work

Along Martin Luther King Jr. Way South in Seattle, collisions and close calls between people walking, biking, driving, and riding light rail happen far too often, largely because of heavy traffic and the many places where trains and vehicles cross at street level. Today, safety improvements usually happen only after several years of reviewing past collision reports, which means changes come slowly and don’t fully reflect what’s causing the problems on the ground. At the same time, the traffic signals rely on older sensors that can only detect some types of vehicles, making it harder to keep all travelers including those on foot, bike, car, or train, moving safely and efficiently.

With advancements in Artificial Intelligence-enhanced video analytics, could this technology be used to identify problem areas and provide a faster pathway for safety improvements?

This pilot project will evaluate the effectiveness of AI-enabled video technology to determine the feasibility for wider deployment to improve safety for all users near at-grade light rail crossings.

For this evaluation, the UW project team developed a process to “ground truth” near-miss incidents identified by the AI-supported video equipment. This application can be used for additional validation efforts to ensure detection systems provide accurate results

The findings from this phase may lead to a wider deployment of safety technology; that may also not only detect patterns that require intervention for improvement but also lead to alerts to users before an incident may occur.

This work is support by Sound Transit, and the Seattle Department of Transportation, funded through a US Department of Transportation SMART Grant.

Civil and Environmental Engineering 

• • Soheil Keshavarz, Research Assistant 

• • Bart Treece, Director, Mobility Innovation Center