Design of Conversational AI Agent Appearance and Trigger

-Focusing on Proactive Information Proposals Triggered by User Behavior and Face Recognition-

Abstract

In recent years, conversational voice support systems such as Siri, Alexa, and Google, which are voice assistants using machine learning technology, have become popular. This is a common function in smart speakers, wearable devices, mobile devices, and PCs. However, when the user requests information provision, it is necessary to call the name to put it in the voice input standby state. In this research, we tried to make a mechanism to identify and recognize the user from the system side and provide information at the timing according to the situation. The system is that recognizing the user’s situation, and then entering the input standby state actively. Also, it is a mechanism to send information according to the situation. This system is realized by combining a general-purpose LiDAR sensor, which is advantageous in terms of cost, and a WEB camera, and using user behavior judgment and face recognition. In the future, it is expected to be used for schedule management by voice, home trainers, and voice learning support linked with ICT teaching materials.