Research on A Multimodal Fusion Intention Understanding Algorithm for Geriatric Carer Robots

In the large application background of man-machine collaboration, the elderly are easy to lose direction or can’t understand the map. When aging users express their intentions to robots, due to ambiguous language or incorrect gestures, robots may have a variety of intentions to understand and analyze. In this paper, an optimization algorithm model is proposed to avoid robot misunderstanding, which can choose the best intention among various kinds of intention understanding, and perform user feedback evaluation on the output intention, so as to understand the real intention of the user. In view of the traditional serial analysis of user behavior in multimodal fusion algorithm limitations, this paper puts forward the algorithm of model through voice recognition and image recognition input information user behavior, care layer parallel fusion user intent from the voice channel and the visual channel, use the feedback mechanism to determine the true intentions of the user from the user’s intent to complete robot navigation links.