Simulation of Dynamic Traffic Light Setting Using Adaptive Neuro-Fuzzy Inference System (ANFIS)

Abstract

Traffic congestion is a common problem in so many big cities. The occurrence of traffic congestion can be caused by various reason, one of them is the increasing number of vehicles while the available facilities are not worth the amount. Congestion can cause many losses, including the amount of time wasted and spent fuel in vain. To solve the problem of congestion can be solved by building a new road. But the solution will make the environment more crowded and require more operational costs. Due to this reason, there is a good solution given that is, the renewal of traffic control system by building intelligent traffic light control system based on road density at each traffic intersection and the width of the road. The method used in this research is Adaptive Neuro-Fuzzy Inference System (ANFIS), this method will process the data of road density and road width to determine the duration of green light to be given. The road density will be calculated based on the video that will be captured by the system and will go through the pre-processing and processing stages until the labelling stage of the object on the road using the Connected Component Labelling (CCL) method, the image pixel of this CCL result will be compared to the image of the empty road condition (no vehicles) to get the percentage of road density. The system testing process used 36 images captured from the video and obtained the accuracy of 98,6%.