딥러닝을 이용한 자율주행차와 일반차 혼재 상황에서의 단기 교통 흐름 예측

Abstract

Intelligent Transportation Systems (ITS) are rapidly developing to improve traffic efficiency, convenience, and safety, and are expected to serve as the foundation for fully autonomous driving in the future. Research on signal policies that consider driving patterns of autonomous vehicles is particularly attracting attention as a key factor in enabling autonomous driving in cities. Because the driving patterns of autonomous vehicles differ from those of human-driving vehicles, a signal cycle suitable for the mixed situation of autonomous vehicles and human-driving vehicles must be derived, requiring highly accurate prediction of traffic volume at intersections. To achieve this, a traffic simulation tool was used to implement a multi-intersection environment in which autonomous and human-driving vehicles are mixed, and a deep learning model based on 1D Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) was developed to predict traffic flow.