Prediction of Switching Times of Traffic Actuated Signal Controls Using Support Vector Machines

Abstract

At signalized intersections there is a significant saving potential of emissions by an energy-efficient and fuel-optimized approach to the stop line. For this purpose, various assistance systems have already been developed. Among other things these systems provide the driver with speed recommendations to cross the next traffic light without stopping. However, accurate information about forthcoming traffic signal switching times is required. Modern traffic signal systems adapt their switching times depending on the current traffic flow. So a predicted phase transition will only occur with a smaller probability than 100%. The paper identifies specific challenges by developing an algorithm for a prediction of traffic actuated signal controls and it presents its mathematical foundations and the results of the prediction.