Vehicular localization using mobile devices for driver assistance applications

Abstract

The ability to accurately localize vehicles is essential for driver assistance applications. Existing modern localization systems, while sophisticated, often come with considerable costs and limitations in accessibility. Recognizing the need for a more accessible solution, there is a growing demand for userfriendly and cost-effective alternatives. One such alternative lies in leveraging mobile devices for localization. However, relying solely on Global Positioning System (GPS), Wi-Fi, or cellular networks, the most commonly used technologies in mobile device localization, often suffer from issues with accuracy. Addressing these drawbacks is crucial to ensuring its effectiveness and widespread adoption. Map matching is a commonly pursued and well-researched approach taken to rectify these erroneous location data. This paper presents the implementation of a hybrid of two map-matching implementations on a mobile device with minimal reliance on external sensory devices: an on-device mapmatching algorithm that utilizes a customizable underlying map when internet connectivity is unavailable, and an open-source Valhalla routing service when the internet is available. This hybrid localization model can be utilized in many driver-assistance mobile applications that require precise vehicle location and has shown promising results during experimentation.