This article presents a proposal for applying neural networks to control road traffic. The proposed solution makes it possible to determine durations of traffic signals at intersections so that the waiting time for transit is as short as possible. The variability of traffic intensity on all access roads and between analysed intersections was taken into account. The developed concept was compared with a method of determining the durations of lights based on the coefficient of intersection readiness, and the feasibility for practical applications of the method was assessed.
