Automatic trip-separation method using sensor data continuously collected by smartphone

A novel method for separating trip periods from non-trip periods, called a “trip-separation method” hereafter, is proposed. Previous studies on such automatic “trip separation” use a fixed threshold to detect “stay periods” (i.e., periods in which a subject is staying put) and separate them from trip periods. However, this separation procedure does not always work well because the collected GPS-positioning data can fluctuate according to the surrounding environment even when the subject stays within an area. The proposed method dynamically adjusts parameters for detecting stay periods according to the surrounding environment. The experimental evaluation on the method showed precision of 81%. This is a promising result implying that sensor-based travel-behavior surveys are possible by utilizing the proposed smartphone-based method.