TranSEC delivers real-time estimations of street-level traffic that engineers can use to manage citywide traffic flow.
To help urban transit managers get a better handle traffic patterns in their cities, the Pacific Northwest National Laboratory (PNNL) has developed a machine learning algorithm that uses publicly available traffic data to predict street-level traffic flow.
The transportation state estimation capability algorithm, or TranSEC, combines data from traffic sensors, the road networks from OpenStreetMaps and Uber Movement, the company’s service that shares mobility data in cities where it operates. Although this data can be sparse and incomplete, TranSEC uses time and trip sampling programs to deliver real-time estimations of street-level traffic that engineers can use to manage overall traffic flow, not just relieve specific bottlenecks.
Using data from the entire 1,500-square-mile Los Angeles metropolitan area, the team leveraged a 64-node, 1280-core supercomputer at PNNL to reduce the time needed to create a traffic congestion model from hours to minutes and visualize the temporal traffic trends at the street-level. TranSEC makes near-real-time traffic analysis feasible, PNNL officials said in a statement.
As more data is collected, the graph-based model, optimization engines and machine learning features of TranSEC evaluates the travel times and routes and refines its analysis of how roadway incidents impact traffic overall, allowing officials to devise mitigation strategies.
Users can update TranSEC with real-time data, and the system also accommodates GPS-based data, as well as weather forecasts or other conditions that could affect traffic.
“What’s novel here is the street level estimation over a large metropolitan area,” said Arif Khan, a PNNL computer scientist who helped develop TranSEC. “And unlike other models that only work in one specific metro area, our tool is portable and can be applied to any urban area where aggregated traffic data is available,” including other modes of transportation, such as freight traffic.
While running a full-scale city model still requires high-performance computing resources, TranSEC can be scaled back for selected use cases in cities where detailed traffic data is available. Only the major highways of an urban road network, for example, could be modeled on a powerful desktop computer, PNNL officials said.
“Traffic engineers nationwide have not had a tool to give them anywhere near real-time estimation of transportation network states,” said Robert Rallo, PNNL computer scientist and principal investigator on the TranSEC project. “Being able to predict conditions an hour or more ahead would be very valuable, to know where the blockages are going to be.”
The team is working to make TranSEC available to municipalities across the country. Eventually, it could also be used to help program autonomous vehicle routes, according to the research team.
Read the full research paper here.