The present invention provides a system and a method to incorporate environmental and personal context 5 to fuse between driver's gaze direction 3 and vehicle location 2 for spoken intent resolution that relates to POIs (Places of interest). We are focusing on the problem of multi-modal (speech & gaze) intent resolution in an in-vehicle scenario for queries that relate to POIs. For example, addressing the query: "What are the opening hours of this restaurant?", and resolving which restaurant the user is referring to (POI resolution 10). Our method uses environmental and personal context as a decision parameter for whether to use gaze d irection or vehicle location for better estimation of POI location. By improving the location estimation, we improve also the intent resolution of the POI. The system: Estimates the type of POI 6 (e.g., "restaurant") in the speech query 1 using an NLU system. Estimates the location of POI 7 according to gaze direction 3 and driving direction 4.
展开▼
