Search

• Estimate link travel time with large-scale incomplete vehicle trip data. • Propose a path-recovery-based method to iteratively estimate link travel time. • Conduct extensive experiments on two public large-scale vehicle trip datasets. The widespread use of positioning devices has generated large-scale trip data, boosting the study of traffic modeling. For instance, New York City Taxi & Limousine Commission regularly releases over 165 million taxi trip records containing the end-point information of each trip every year. Such big datasets provide us potential new perspectives to tackle traditional issues in traffic modeling. In this paper, we propose to use such vehicle trip data, which do not contain information on the intermediate vehicle positions other than the end points, to study the link (road segment) travel time estimation problem. A method named PR-LTTE is developed to estimate the mean travel time for interested links in a road network based on path recovery, jointly modeling trip distance and trip travel time. The key idea is to iteratively alternate between 1) inferring the most likely path for a given trip by minimizing a loss function that considers both trip distance and trip travel time, and 2) computing the link travel time via the least-squares estimation with the inferred path. Our experiments on two large-scale trip datasets (including New York City taxi trip data and Chengdu DiDi trip data) show very promising results. PR-LTTE improves the accuracy of travel time estimation by more than 50% in most cases over state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Exploiting deep learning techniques for traffic flow prediction has become increasingly widespread. Most existing studies combine CNN or GCN with recurrent neural network to extract the spatio-temporal features in traffic networks. The traffic networks can be naturally modeled as graphs which are effective to capture the topology and spatial correlations among road links. The issue is that the traffic network is dynamic due to the continuous changing of the traffic environment. Compared with the static graph, the dynamic graph can better reflect the spatio-temporal features of the traffic network. However, in practical applications, due to the limited accuracy and timeliness of data, it is hard to generate graph structures through frequent statistical data. Therefore, it is necessary to design a method to overcome data defects in traffic flow prediction. In this paper, we propose a long-term traffic flow prediction method based on dynamic graphs. The traffic network is modeled by dynamic traffic flow probability graphs, and graph convolution is performed on the dynamic graphs to learn spatial features, which are then combined with LSTM units to learn temporal features. In particular, we further propose to use graph convolutional policy network based on reinforcement learning to generate dynamic graphs when the dynamic graphs are incomplete due to the data sparsity i sue. By testing our method on city-bike data in New York City, it demonstrates that our model can achieve stable and effective long-term predictions of traffic flow, and can reduce the impact of data defects on prediction results. [ABSTRACT FROM AUTHOR]