Travel surveys form a key component of transportation planning by making transportation related data available to the planners. Smartphones are emerging as ideal tools for collecting detailed individual travel information, which motivated the authors to develop a smartphone-based travel survey system, Future Mobility Survey (FMS). Inferring people’s stops and modes of transportation is a critical and challenging problem in the FMS system, especially because time phased data collection method has been used to reduce the battery usage. In this paper, the authors propose a novel algorithm for integrated stop and travel mode detection using parsimonious real world data collected from smartphones through FMS. The authors use a two stage classification system to detect five modes of travel, viz., stop, walk, train, car, and bus. To improve accuracy of the classification the authors derive features from a fusion of data collected from GPS,GSM,Wi-Fi and Accelerometer sensors on-board the smartphones. The authors also propose new features based on contextual information and user’s historical data. Experimental results show that the algorithm can effectively perform stop/mode detection and is robust against noisy/incomplete data from the smartphones.
Stored only when both papers are in the corpus — 14 citations exist upstream.