Abstract
The present disclosure describes systems and methods that use activity recognition detection (e.g., step detection) and machine-learned course prediction models to generate user path approximations from sensor data obtained by a user computing device, which is particularly beneficial when GPS devices are unavailable. User paths generated from a plurality of different mobile computing devices can be utilized collectively to build maps of walkable spaces. Keywords associated with the present disclosure include: mapping; locations; walkable spaces; machine learning; model; training; GPS; sensors; activity recognition; sensor fusion; crowdsourcing; collaborative learning; user path; path generation; course prediction.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
N/A, "Mapping Walkable Spaces Using Activity Recognition", Technical Disclosure Commons, (January 30, 2018)
https://www.tdcommons.org/dpubs_series/1034