Overcoming battery life problems of smartphones when creating automated travel diaries

Gathering data using travel diaries has been a requirement to analyze travel behaviour data for decades. Traditional pen and paper techniques of gathering data has been complemented by new technologies, most recently GPS loggers, and now smartphones. Smartphones appear to be the silver bullet in collecting travel data on a wide scale, they are widespread, applications are easy to distribute, and the hardware is reliable. However, conserving the battery life in smartphones is a challenge to creating automated travel diaries. The smartphone is not solely a travel diary tool and for a system to work, the travel diary application cannot be a burden to the user. In this paper we leverage work from the “everyday location monitoring” research community in a system to sparsely collect location data from smartphones in a battery efficient manner. With this sparse data, we present an algorithm to generate trips, and suggest six metrics to analyze the quality of the trip determination system. We show how the algorithm has been adapted to deal with problems encountered in the real world (i.e. inaccurate location readings, different capabilities of various phones, various user behaviours turning phone services on/off), with a test of 125 users across 20 states and 23 different models of smartphones. We also evaluate the system across the six defined metrics with a three-month dataset from six users. The results show the accuracy of the automated system, and how the results are significantly boosted with just a small amount of human input. The results show that there exists a trade-off between battery consumption and trip inference accuracy, and the metrics used in this paper ar