Search Results

Intelligent Transportation System (ITS) plays an important role in smart city, and accurate short-term traffic flow prediction is a significant part. At present, China’s ITS has developed rapidly, and advanced intelligent transportation systems have been built in major cities, such as Shanghai, Shenzhen and so on. With the promotion of mixed Electronic Toll Collection (ETC) and Manual Toll Collection (MTC) charging systems, the features of the traffic flow data have become richer. Traffic data recorded some information for the vehicles entering and exiting highway toll station including time, location, type, mileage, then we can use historical OD data to do traffic flow prediction, predict the corresponding future exit station traffic flow. Furthermore, due to the deep learning network’s ability to model deep complex non-linear relationship in data, researchers have paid more attention to predict traffic flow using deep learning models in recent years.

Urban intersection is the key element to determine the traffic operation of road network. Under the CAVs environment, the roadside control equipment of intersection can communicate with CAVs in real time, collect vehicle state data and optimize traffic control schemes. This paper presents a method for intersection traffic signal control based on deep learning of CAVs data. In addition, intelligent control agent of traffic signal (ICATS) is designed to simulate CAVs. ICATS can perceive real-time changes of traffic flow, model different conditions of intersection and generate the corresponding traffic signal scheme. ICATS used double Q-learning method combination with deep neural network, which is an effective model-independent deep learning algorithm. Moreover, the real traffic data is collected and tested in this paper for evaluating the experiment performance, including vehicle delay, number of passing vehicles, total stop times and passing time.

The recent research on location approaches of the intelligent vehicle based on Light Detection and Ranging (LiDAR) is analyzed in this paper. According to the features of these approaches, it can be divided into three categories: simultaneous localization and mapping (SLAM), offline mapping and online localization (OMOL) and fusion localization (FL). Past research and applications of the main algorithms and critical research scenarios in each localization approaches are reviewed. Three aspects of the current trend in location approaches of the intelligent vehicle based on LiDAR are discussed. Based on object detection, object recognition and object analysis algorithms in the field of deep learning, semantic SLAM and real-time three-dimensional reconstruction are important research trends for SLAM. The performance of robustness and real-time performance of localization algorithm of intelligent vehicles based on LiDAR need to be improved.