Real-time adaptive signaling for isolated intersections

Abstract

Efficient design is required for a cyber-physical system, as we need to trade off the complexity and performance benefits. We consider the CPS of a traffic-light controller at an isolated intersection that is used by autonomous, semi-autonomous, and human-driven automobiles. Existing traffic systems are vulnerable to accidents (more than 1 million people die in automotive accidents globally) and undesired traffic delays (the average U.S. driver spends a week stuck in traffic every year). The next generation of traffic-light control systems should protect against these disruptions while maintaining enhanced control of the system to optimize features of interest. The potential benefits of such a system include increased safety in the presence of higher density traffic, increased fuel and time efficiency (as less time is wasted in queuing), and decreased demands on drivers to make driving decisions. Here, we demonstrate a novel approach to adjusting the cycle length, yellow time, and red-to-green ratios of a traffic signal by minimizing the average loss per vehicle due the presence of the signal.