Robot behavior generation method

The purpose of the present invention is to provide a robot behavior generation method that more accurately reproduces human behavior. To accomplish this purpose, there is provided a robot behavior generation method for the purpose of reproducing human behavior with a robot, wherein the robot behavior generation method has: a step S1 for generating a human body model by modeling the configuration of the human body; a step S2 for generating a robot model by modeling a configuration of a robot; a step S3 for creating associations between the human body model generated in step S1 and the robot model generated in step S2; a step S4 for acquiring first movement data indicating displacement of a human body model corresponding to behavior of the human body; a step S5 for selectively choosing movement feature indicators having first unknown movement data that indicates displacement of the human body model generated in step S1, and formulating evaluation criteria pertaining to reproduction error, for normative movement data acquired in step S4; a step S6 for selecting a constraint necessary for the purpose of movement by a robot which has second unknown movement data that indicates displacement of the robot model generated in step S2; a step S7 for calculating first unknown movement data and second unknown movement data such that reproduction error based on the evaluation criteria of step S5 is minimized under the constraint selected in step S6 and the association obtained in step S3; and a step S8 for using the second unknown movement data calculated in step S7 to control the robot.