Who you are is how you travel: Detecting mode of travel from people's movement patterns

They enable people to carry what would not so long ago have been a powerful desktop PC around in their pocket, opening up unprecedented opportunities for communication, navigation, entertainment and business on the move. This rise in smartphones has led to the generation of vast amounts of data on the movements and activities of individuals. Used responsibly, such data can reveal fascinating insights into the collective behaviour of people. Google’s live traffic information is one such example.

The problem with such data is they tend to be ‘noisy’ and ‘unlabelled’ – we know broadly where people have travelled, but not how they travelled, where they were going precisely, or what the purpose of their trip was. At UCL, we are researching how to use smartphone data to infer both the type of transport people use as they move about the city, and the types of activities they carry out when they stop at a location.

In particular, we are interested in how patterns in the data differ in people with different mobility impairments. This involves integrating data from smartphones with individuals’ characteristics and geographic data from other sources, and feeding them into a Bayesian probabilistic model. The latest progress in this research appears in our paper, ‘Who you are is how you travel: A framework for transportation mode detection using individual and environmental characteristics’, published in Transportation Research Part C: Emerging Technologies (July 2017).

This research makes use of the data generated as people use their smartphones to learn about how people move around cities, with emphasis placed on those with mobility impairments. A technique has been developed to automatically detect what transport mode people are using based on their movement traces.