MateBot: The Design of a Human-Like, Context-Sensitive Virtual Bot for Harmonious Human-Computer Interaction

The virtual bot is one of the hot topics in artificial intelligence, where most of the current studies focus on chatbots. Nevertheless, the context-sensitive virtual bot, especially with rich human-like interactions (e.g., appearance change, context-aware narration/recommendation) regarding the ambient changes (e.g., location, focused scene) through various built-in sensors, would have broader application. Towards this direction, we propose MateBot, a human-like, context-sensitive virtual bot, which supports harmonious human-computer interaction on smartphones. The design of MateBot consists of three parts. First, a context sensing network is used to recognize the input background information and face information, and modify the appearance of the virtual bot through the conversion of the encoding network. Second, a human-like bot appearance generation network can generate a virtual bot image with a human-like appearance through the GAN network and modify the appearance of the virtual bot with context-sensitive information. Third, a personalized conversation network is devised to communicate with human users. Furthermore, we apply MateBot to the intelligent travel scenario to justify its practicality, and the experiment results show that the bot can better increase the user’s sense of substitution and improve the communication efficiency between human users and virtual bots.