Exploitation of Context Information for Natural Speech Dialogue Management in Car Environments

Abstract

This chapter focuses on the exploitation of context information for a situation- and user-aware dialogue management implemented in a framework for the automotive environment. A generic dialogue manager for driver information systems (like infotainment and communication systems) and driver assistance systems has been developed and tested. One main focus in the development was the ability to make context-dependent decisions. The dialogue manager provides flexible and user-centered speech dialogues and supports multimodal interfaces, like buttons or turning knobs combined with speech. A frame-based approach is used for the dialogue control. The XML description allows an easy specification and overview over the dialogue structure. Visual outputs are monitored on several displays in the car. The evaluation shows an improvement of effectiveness and a higher joy of use through the possibility of submitting several pieces of information in only one dialogue step with natural speech comparing to a menu-based spoken dialogue. The context-dependent information agents reached a high acceptance by the users. The test persons rated the context-based way of frame-based interaction as comfortable and important.