Volvo intelligent news: A context aware multi modal proactive recommender system for in-vehicle use

Abstract

Keeping up to date with the world outside while driving is currently only possible through the impersonal car radio or the distracting features of a smart phone. In-vehicle infotainment systems exist today that attempt to tackle this challenge but few are sensitive to the current driving state. This paper describes the development of a multi-modal proactive recommendation system featuring personalized content combined with the use of car sensors to determine when to present information. A prototype is implemented in a car simulator and trained through user tests. The final system shows a 19% deviation from baseline driving and thereby performs better than comparable systems. This is backed up by user experience data which reports drivers having a positive experience with the system.