Abstract
Mobile navigation applications for public transit can experience degraded performance in environments with poor global positioning system signals, such as subway tunnels, which may cause a user's location display to freeze and guidance to cease. A described on-device system may address this by fusing data from a plurality of sensors, such as a barometer, magnetometer, and accelerometer, on a computing device like a smartphone or wearable device. A lightweight machine learning model can process this sensor data to generate a real-time probability that the user is on a moving transit vehicle. This probability may then be combined with a user's planned itinerary via a contextual matching algorithm to infer semantic events like boarding or alighting. This method can enable more continuous progress tracking along a pre-defined transit route in signal-denied environments, allowing a navigation application to provide more consistent guidance and a responsive map display.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Shi, Hui Ling; Lai, Louis; Dolk, Oliver; Kamisli Hasanoglu, Gul; and Ong, Si Min, "On-Device Multi-Sensor Fusion with Machine Learning for Continuous Transit Navigation", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/9977