Twitter
Google plus
Facebook
Vimeo
Pinterest

Fluid Edge Themes

AI Helps Predict Traffic Jams

Home  /  News   /  AI Helps Predict Traffic Jams
LA Traffic Eased With Artificial Intelligence

AI Helps Predict Traffic Jams

AI Helps Predict Traffic Jams

L.A. traffic is a thing of legends, and for good reason. Angelinos spend an estimated 120 hours a year stuck in it. In order to better understand and predict traffic, researchers at the Argonne National Laboratory collected nearly a year’s worth of data from over 11,000 traffic sensors throughout the L.A. area and fed it to an AI model.

LA Traffic Eased With Artificial Intelligence

With all that information stored and a supercomputer at their disposal, the system can look at current traffic trends and predict the next hour of traffic with extreme accuracy in milliseconds. This speed and accuracy is made possible by a technique called graph-based deep learning, which allows near automatic training and ongoing machine learning.

With over 11,000 highway sensors updating in real time, even supercomputers can struggle to keep up. But the new graph-partioning method implemented at the Argonne Center for Transportation Research allowed all this data to be processed in near real time, resulting in a predicted speed accuracy of within 6mph of real traffic speed, one hour in the future.

This data can now be utilized by traffic management departments, allowing quicker response to a potential traffic jam, and could even be used for planning infrastructure improvements across the L.A. area.