Özet:

Traffic accidents are substantially caused by driver faults and breaking the rules. With the studies conducted in recent years, developments in electric and autonomous vehicle technologies are progressing very rapidly. By the systems developed, it is aimed to prevent the driver-induced accidents. The main problem in autonomous vehicles is the recognition of traffic signs and markers while running in real time. The data from the sensors and cameras used in autonomous vehicles are transformed into meaningful results with established algorithms. In this way, it is aimed that the vehicle will act in accordance with the traffic rules, independent of the driver. Within the scope of this study, the data of traffic signs and markers used in Turkey were collected and different data sets of six signs were created. Using these data sets, trainings were carried out via Haarcascade machine learning algorithm. The traffic signs trained via the Haarcascade method were defined on the track created in a three-dimensional simulation environment. Training data were tested by simulating the behavior of the autonomous vehicle on the track. In the simulation environment, the motion control of the autonomous vehicle was successfully performed by this data.