Akıllı Ulaşım Sistemleri ve Uygulamaları Dergisi

Yazarlar: İlgin GÖKAŞAR, Selim DÜNDAR

Konular:-

Anahtar Kelimeler:Autonomous vehicles,Intelligent transportation systems,Machine learning,Traffic simulation

Özet: In recent decades, autonomous vehicles are introduced as a result of significant developments in artificial intelligence. In near feature, it is possible that more and more autonomous cars will be a part of the daily traffic. Autonomous vehicles can communicate with other vehicles, so they have much lower response time than the human drivers, thus, autonomous vehicles can be operated with lower headway. This feature is expected to significantly increase the capacity of roads when completely autonomous vehicles are operated in traffic and to reduce traffic accidents. However, the effect is more complicated in combined traffic conditions where autonomous and human-driven vehicles are present. Studies have shown that autonomous vehicles adversely affect traffic flow characteristics in combined traffic conditions. In this study, the effects of autonomous vehicles on traffic network is evaluated using 15 scenarios including different traffic demand levels and autonomous vehicle composition on a microsimulation model of a network in Turgut Özal (Millet) Caddesi in Istanbul. Then, the effect of the traffic parameters is analyzed and predicted using Artificial Neural Networks. Artificial Neural Network model is capable of estimating the average flow, speed and acceleration with a higher accuracy than average headway, gap and occupancy.

---