Driver Identification Based on Gear Shift Events and Attention-Based Bidirectional Long Short-Term Memory for Manual Transmission System

The rapid development of Internet of Vehicles (IoV) and in-vehicle interconnection systems makes it possible to obtain massive, real-time, and multi-dimensional driving behavior data. In this paper, we present a new method for driver identification based on gear shift events which are extracted from naturalistic driving data. To tackle the problem of sequence classification, the Attention-based bidirectional long short-term memory (Bi-LSTM) algorithm is adopted. In the experiments, over 15000 gear shift events of each driver among 20 drivers are selected to analyze and model. Experiments show that our method has the best performance to identify drivers with an improvement of 10% compared to methods using continuous driving data. Moreover, our method can extract hidden features from driving behavior data time-efficiently without complicated feature preprocessing and engineering.