Generation Method of Evaluation for a Robot Considering Relations with Other Robots

Abstract

This paper proposes a method of generating behavioral evaluations that considers relations to other robots to enable a single operator to control the behavior of multiple robots. We previously proposed an algorithm for individual robots to select their behavior considering the behavior of other robots, using a simple signal given by an operator to the group of robots as a whole. In this algorithm, a learning function was installed on individual robots to enable them to avoid conflict with other robots, and the behavioral evaluation necessary for executing learning had to be given to individual robots based on the situation. This paper explains how individual robots correct their own behavioral evaluation based on the signal given to the group by the operator. We propose behavioral evaluation correction that takes into account cases in which the operator makes mistakes in sending signals to robots or forgets to do so. We verify the effectiveness of the proposed method by simulation.