Deep learning based urban vehicle trajectory analytics

Abstract

A `trajectory' refers to a trace generated by a moving object in geographical spaces, usually represented by of a series of chronologically ordered points, where each point consists of a geo-spatial coordinate set and a timestamp. Rapid advancements in location sensing and wireless communication technology enabled us to collect and store a massive amount of trajectory data. As a result, many researchers use trajectory data to analyze mobility of various moving objects. In this dissertation, we focus on the `urban vehicle trajectory,' which refers to trajectories of vehicles in urban traffic networks, and we focus on `urban vehicle trajectory analytics.' The urban vehicle trajectory analytics offers unprecedented opportunities to understand vehicle movement patterns in urban traffic networks including both user-centric travel experiences and system-wide spatiotemporal patterns. The spatiotemporal features of urban vehicle trajectory data are structurally correlated with each other, and consequently, many previous researchers used various methods to understand this structure. Especially, deep-learning models are getting attentions of many researchers due to its powerful function approximation and feature representation abilities. As a result, the objective of this dissertation is to develop deep-learning based models for urban vehicle trajectory analytics to better understand the mobility patterns of urban traffic networks. Particularly, this dissertation focuses on two research topics, which has high necessity, importance and applicability: Next Location Prediction, and Synthetic Trajectory Generation. In next location prediction, we propose deep-learning based models that considers the spatiotemporal patterns of urban vehicle trajectories. First, we partition the urban traffic network into cells and represented urban vehicle trajectories as cell sequences to extract the spatial features from trajectory data. In addition, we used recurrent neural network (RNN) to predict the next location. Furthermore, to improve the performance of RNN model, we propose attention-based recurrent neural network (ARNN) model, which incorporates the network-wide traffic state information into next location prediction. The performance of the model is evaluated in both aggregated region level and individual trajectory level, and the proposed model has better performance than the baseline model. In synthetic trajectory generation, we propose TrajGAIL (Generative Adversarial Imitation Learning for Urban Vehicle Trajectory), which reproduce trajectories with both patterns as individual trajectory and patterns as a group. TrajGAIL uses the Generative Adversarial Imitation Learning (GAIL) and Partially Observable Markov Decision Process (POMDP) to understand the spatiotemporal patterns and reproduce realistic trajectories. The performance of the model is evaluated in trajectory-level and dataset-level, and the results show that the proposed TrajGAIL shows an outstanding performance compared to the baseline models. In this study, we propose various novel models for urban vehicle trajectory analytics using deep learning. In three different research topics, we analyzed the current challenges in each topic, propose research approaches to resolve the challenge, and developed a novel model based on the research approaches. By using the proposed model, it is expected to increase the applicability of urban vehicle trajectories in various fields of study.