Path2Go: Context-aware services for mobile real-time multimodal traveler information

Abstract

Mobile platforms are now becoming a more and more important medium for providing information to travelers on the move. To improve the accuracy and relevance of the mobile traveler information, context awareness has become a active research topic. In this paper, we describe algorithms and services provided by Path2Go, a multimodal traveler information system developed by California PATH, UC Berkeley. The Path2Go activity detection algorithm, the core of the context-aware design, has made the following contributions: (1) it uses a rule-based multi-hypothesis Bayesian method for mode detection, to address the limitations of using mobile phone GPS for activity detection and increase the speed of convergence; (2) it fuses GPS data from transit vehicles and the GPS of the user's mobile phone for better activity detection; and (3) it enables several experimental services, including variable-frequency client-server communication and need-based GPS use. Field testing of the Path2Go activity detection algorithms showed reasonably good results. The Path2Go application has also been made available to the public through iPhone, Android and Windows Mobile platforms, with some of the features discussed in this paper included.