Over the past four years researchers in the University of Washington (UW) ACCESS project have developed machine learning and ubiquitous computing algorithms that support new kinds of intelligent wayfinding systems. This work includes methods for (i) inferring a user's mode of transportation (such as walking, bicycling, or riding on a bus) in order to provide mode-appropriate information; (ii) learning multi-step transportation plans by demonstration, rather than by manually entering a complex route; (iii) learning significant places in a user's life (such as the user's home, workplace, and friends homes) without requiring explicit user input; (iv) predicting the user's most likely destination based on the user's current location and the time of day, so that the user can simply confirm a destination rather than manually typing it in; and (v) detecting unusual variations from a user's typical daily movements, that may indicate the user is lost and needs help.
Phase I of the collaboration between Sendero Group, UW and University of Rochester investigated methods for implementing and deploying the 5 functions described above on a robust, commercial personal GPS platform. In Phase I, the mode of transport function (function (i), above) will be fully implemented and deployed in a prototype using the existing Sendero GPS platform.
Michael May, Principal Investigator
Charles LaPierre, CTO
Henry Kautz, Ph.D. Department of Computer Science at the University of Rochester
>Kurt Johnson , Ph. D. Professor in the School of Medicine at the University of Washington where he serves as head of the Division of Rehabilitation Counseling in the Department of Rehabilitation Medicine.
Gaetano Borriello, Ph.D. Professor
Gil Lutz Staff Tester