Exploiting Multi-source Data for Adversarial Driving Style Representation Learning

Abstract

Characterizing human driver’s driving behaviors from GPS trajectories is an important yet challenging trajectory mining task. Previous works heavily rely on high-quality GPS data to learn such driving style representations through deep neural networks. However, they have overlooked the driving contexts that greatly govern drivers’ driving activities and the data sparsity issue of practical GPS trajectories collected at a low-sampling rate. To address the limitations of existing works, we present an adversarial driving style representation learning approach, named $\mathtt {Radar}$. In addition to summarizing statistic features from raw GPS data, $\mathtt {Radar}$ also extracts contextual features from three aspects of road condition, geographic semantic, and traffic condition. We further exploit the advanced semi-supervised generative adversarial networks to construct our learning model. By jointly considering statistic features and contextual features, the trained model is able to efficiently learn driving style representations even from sparse trajectories. Experiments on two benchmark applications, i.e., driver number estimation and driver identification, with a large real-world GPS trajectory dataset demonstrate that $\mathtt {Radar}$ can outperform the state-of-the-art approaches by learning more effective and accurate driving style representations.