IMPROVING DECISIONS OF AN INTELLIGENT SEMAPHORE

This work is dedicated to compare and evaluate the performance of a CBR approach to control intelligent semaphores using a fuzzy inference system and an inductive decision tree approach to decide when the green light of a specific semaphore must be modified. Such decision is relevant in order to achieve a suitable vehicular mobility service level in urban environments. Derived from previous works, we argue that a traffic light signal can improve the service level of a specific junction if it is capable of fixing the duration of the green light interval in base on the road conditions. However, the duration of such light should not be modified at every cycle. The semaphore must evaluate if it is necessary to change the green interval before to develop a more complex analysis. To do this, the proposal is to include a process in which the semaphore reviews some road attributes to take a suitable and trustworthy decision. The fuzzy logic control and ID3 approaches are methods implemented to perform support decision under a dynamic and unpredictable set of information. Some experimental tests show promising results when the traffic devices support their decisions over the indicated techniques. For example, using the ID3 approach the decisions of the system improves almost 27% in a set of 10,000 experiments. Meanwhile using a fuzzy logic control the performances increase in a 20% in the same amount of proves. Finally, some conclusions are drawn to emphasize the benefits of including these techniques in a methodology to implement intelligent semaphores.