DALI : an Autonomous, collaborative AI-based Traffic Control System

Problem:

The impact of traffic congestion on individual driver's time is well known with US residents spending an average 99 hours annually in gridlocks. A study by INRIX indicates that traffic congestion robs U.S. households of $1,377 and the U.S. economy of $88 billion annually. The report also reveals that the cost of traffic congestion is projected to reach $2.8 trillion in 2030. In addition to travel costs, the increasing severity of congestion has a direct impact on the environment. High levels of pollutant emissions degrade air quality and contribute to higher risks of morbidity and mortality for drivers and individuals living near roadways.

At the core of the traffic congestion problem is the inadequate performance of traffic signal systems which still operate according to 20th century standards.

Solution:

DALI (Distributed Agent-based traffic LIghts) is an innovative, adaptive traffic control system to reduce congestion and improve the driving experience.DALI aims to make existing traffic control systems autonomous and smart without the need for expensive changes or upgrades. We achieve this goal by plugging AI software called agent into each existing intersection controller, which becomes "the brain" of the controller. The agents analyze the traffic data, communicate directly, and work together to execute a timing strategy that improves traffic flow in real-time. As a plug-in, AI-based software solution, DALI can be scaled out to a city of any size.

DALI was successfully simulated on a digital twin of the City of Richardson and deployed at three intersections. Initial results reflect a 40% reduction in traffic delays. Phase II and III deployment are due to start soon.

 

 

Technology Readiness:

Level 7

Technologies:

Multi-Agent Systems, AI: Planning, Collaborative Decision Making, Machine Learning