AI and Machine Learning Models to Solve Traffic Jams in LA


A group of computer science researchers from the Lawrence Berkeley National Lab are working on machine learning and high performance computing (HPC) with an objective to improve the real-time decision making ability of Caltrans. Caltrans is an acronym for the California Department of Transportation, who is also working with the researchers.

Generally, regular urban traffic follows a timely pattern with respect to a usual 9 to 5 work hours. However, an accident or a mishap on a road may disrupt this pattern. Designing precise traffic flow modules, for their use during such times, is a great challenge for traffic engineers. Moreover, these traffic engineers must consider and adapt to the unforeseen traffic situations in real time.
This has been the main motive behind the on going research by Lawrence Berkeley National Lab and Caltrans. California Partners for Advanced Transportation Technology are also part of this research. Moreover, the Institute for Transportation Studies (ITS) UC Berkeley and Connected Corridors are also working in collaboration for testing, researching, and developing traffic models in California.
Solving Traffic Problems
Connected Corridors and Caltrans are now working on an implementation of a system in LA County with the help of the I-210 pilot. By using real time information, the objective is to enhance the real-time decision making ability of Caltrans. The real-time data will come from partners in southern California at the state, county, and city level. The objective is to enhance the execution of coordination of multijurisdictional traffic. Furthermore, the system also aims to optimize the incident response time to cut down the negative effect of an accident on traffic.
The first round of such a system will go live across the cities of Pasadena, Monrovia, Duarte, and Arcadia in 2020. The next step of the plan is to deploy the system across the state of California.

Leave a Reply