Mode Classification for Mixed Traffic Flow Based on Smartphone Data

Abstract

In Southeast Asian countries, fixed-point sensors for road traffic are under development, while the information and communication infrastructures have rapidly improved. Thus, smartphones with various sensors, such as GPSs and accelerometers, are commonly used. As a result, it is possible that traffic conditions could be monitored in real time using smartphone data rather than fixed-point traffic detectors. We have developed a method to classify transportation modes (i.e., passenger vehicles or motorcycles) using data collected from smartphones. The experimental results indicate that the standard deviation and spectrum of synthetic acceleration are measures that can accurately classify transportation modes.