Highway Driving Events Identification and Classification using Smartphone

Abstract

Research and developments in the newly emerging vehicular applications such as driving monitoring systems, driving behavior and style analysis, driving intension modeling and vehicle telematics, have greatly contributed in the fields of road safety analysis, intelligent transportation systems and microscopic traffic simulation for smart cities. Identification and classification of driving events represents a fundamental necessity for all these systems and in fact they represent the backbone module for any successful application. In recent years, the use of smartphones has grown significantly due to the increase in their computational capabilities and the integration of advanced sensor technologies. This prevalence of smartphones and advances in machine learning techniques have rapidly transformed the field of vehicular applications to be easily accessible, widely available, and implemented at low cost. This paper presents a simple but an effective approach for the identification and classification of driving events. The approach is based on separating events identification process from the classification process. The Dynamic Time Warping (DTW) technique is used for the identification, while statistical and time metrics features are used for the classification. Results obtained show a high accuracy rate of the proposed system.