No abstract

With the advent of wearable GNSS devices and activity trackers, new opportunities for automatic travel-mode detection arise. Although physiological measures such as heart rates carry a high potential for travel-mode detection, little research has been done that exploits this data. This paper presents a rule-based method for the detection of the travel modes walk, bike, bus, train and car, based on the combination of GNSS and heart-rate data from off-the-shelf fitness watches. The aim of this research is to minimize the input variables and reference data for mode detection. In the case study, the proposed workflow performed very well and substantially reduced the confusion between active and motorized travel modes compared to a workflow that did not take heart rate into consideration, although the differentiation among motorized travel modes could be further enhanced with additional data. Combining GNSS data with physiological variables such as heart rate allows a clear reduction in the amount of reference data and processing effort required for mode detection.