It’s no big secret that Los Angeles is infamous for its heavy traffic jams, regularly ranking 1st in studies of the country’s traffic hot spots. Estimates propose that Angelinos spend an extra 120 hours a year stuck in them. While a bad dream for drivers, the Los Angeles transportation system does have its advantages if you’re devising a new system to quickly predict and potentially divert that traffic.
Scientists from across the U.S. Department of Energy’s (DOE) Argonne National Laboratory set out to do exactly that under the umbrella of a bigger project on the design and planning of mobility systems drove by collaborators at DOE’s Lawrence Berkeley National Laboratory (LBNL).
Utilizing an AI method called machine learning, the team utilized Argonne’s supercomputers to process traffic patterns from almost a year’s worth of data taken from 11,160 sensors alongside the enormous California highway system. That data was then used to prepare a model to forecast traffic at lightning-fast speeds—certainly faster than L.A. traffic. Within milliseconds, the model can look at the past hour of information, and forecast the next hour of traffic with great accuracy.
The study published in the Journal of the Transportation Research Board (2020)
Prasanna Balaprakash, a computer scientist in MCS with a joint appointment in the ALCF said, “The AI and supercomputing capabilities that have been used in this work allow us to tackle huge problems.”
“The scale of this project is large, and this amount of data requires an equally large computing resource to tackle it.”
Using ALCF’s world-class computing resources, researchers drastically reduced the number of computer hours needed to train the model. For example, where it might take a top-of-the-line desktop computer a week to train the traffic forecasting model, the same process can be done in three hours on a supercomputer.
Harnessing the power of graph-based deep learning a complex form of machine learning that — can make decisions and improve a model’s predictions almost automatically their model uses historical data to predict traffic patterns while forecasting speed and flow at the same time. This is important because traffic flows in one area at any given time depending upon the speed and flow of traffic nearby.
Eric Rask, a former principal research engineer with Argonne’s Center for Transportation Research and one of the scientists involved in the study said, “Traffic forecasting approaches are critical to developing adaptive strategies for transportation.”
“Traffic patterns have complex spatial and temporal dependencies that make accurate forecasting on large highway networks a challenging task.”
Previous models could handle data from only 200-300 sensor locations; but with this new graph-partitioning method, the team was able to process data from over 11,000 locations while reducing the model training time by an order of magnitude. This method was not only fast, it could accurately predict the speed of traffic an hour into the future, typically within 6 mph of the observed speed at any location across the network.
Tanwi Mallick, a postdoctoral appointee in MCS said, “The scale and accuracy of the forecasting techniques have the potential to enable better decision making.”