SENSg: Large-Scale Deployment of Wearable Sensors for Trip and Transport Mode Logging

Abstract

In this work, the authors present a large-scale individual mobility sensing project based on custom sensor nodes (SENSg) and infrastructure, together with preliminary results and analysis based on three trial runs. Between September and November 2015, 50,000 SENSg nodes will be used by Singaporean students to collect data during their daily activities as part of a National Science Experiment. While the experiment focuses on early technical education for students, the authors present the technical infrastructure for the project, their algorithms to analyze mobility aspects of the collected data, and preliminary results from the pilot trials. In particular, they focus on the goals of 1) measuring student activity through pedometry, 2) quantifying the amount of time students spend indoors or outdoors 3) analyzing student transportation mode, trip origin/destination and travel times. Using pilot data, the authors found their embedded identification algorithms to be more than 75% accurate in determining whether a student is indoors or outdoors, and more than 60% accurate in identifying travel mode. In addition, they analyzed the movement patterns of the participants to gain insights, for example the distance of their homes to their schools. They show that Primary students typically reside much closer to their schools than Secondary students, and the mean commuting times for students are between 14-25 minutes. Future work will examine cycling modes in different cities (bicycles are not common for commuting in Singapore), in a straightforward extension of existing algorithms. The authors will expand their analysis once data from the large scale deployment is available.

Publication
Proceedings of Transportation Research Board Annual Meeting (TRB)