Intelligent Agent Based Traffic Signal Control on Isolated Intersections

The purpose of this paper is to develop an adaptive signal control strategy on isolated urban intersections. An innovative approach to defining the set of states dependent on the actual and primarily observed parameters has been introduced. ?he Q?learning algorithm has been applied. The developed self-learning adaptive signal strategy has been tested on a re?l intersection. The intelligent agent results have been compared to those in cases of fixed-time and actuated control. Regarding the average total delay, the total number of stops and the total throughput, the best results have been obtained for unknown traffic demand and over-capacity.