Action selection of robot by human intention estimated with dynamic evaluation criterion

In recent years, a robot is active in human living space. A robot is also required to perform multiple tasks in general. Moreover, a robot is also required to support a human according to its intention. We proposed a method that is decision-making for the robots with multi-task using human-robot natural interaction. In this method, a degree of achievement request was defined as an indicator that wishes from human to robot for task achievement. The robot selects an action on the basis of the degree of achievement request. Thereby, the robot can take an action according to a human intention. However, previous research did not adapt to individual differences in interactions when determining a degree of achievement request. Because, the interaction was evaluated on the basis of the static evaluation criterion when estimating a human intention. There is a possibility that a robot will determine a degree of achievement request different from a human intention. In order to solve the problem, our proposed method is dynamically changing an evaluation criterion for each human who interacts with a robot. It adapts to individual differences in interaction by dynamic evaluation criterion. Anyway, an experiment to compare the proposed method and the method of previous research was conducted. From experimental results, the usefulness of the proposed method has been demonstrated.