Behavior-based intention inference for intelligent robots cooperating with human

A method for intention inference from the human's behavior is proposed for intelligent robots who carry out a simple task cooperatively with human, without complex communication. The system is composed of three levels; perception, recognition and intention inference. At perception level, robots get the physical information about the environment including human by processing data from sensors. The system translates them into qualitative expression that contain vague time scale, e.g., now, for a while, for all this while, by using fuzzy logic in recognition level. Finally, the intention inference level has groups of fuzzy rules using qualitative expression for the specific cooperative task to infer the human's intention and to eject from the task that is reasoned in intention inference level by fuzzy rules matching the qualitative expression to the specific situations. Intelligent robots recognize the means of human's behavior by the system like human doing, and the system will be of use to construct the human-robot cooperative working system for industry. We set up a simple human-robot cooperative task and ascertain the functions of the system for intention inference corresponding to the case.<<ETX>>