A Study of Human State Pattern Acquisition Method for Partner Robots

Abstract

For a partner robot that supports a human, it is important to know the state pattern of the human in the environment so that the robot can generate personalized motions according to the state. In this study, we studied a human state classification method to enable the system to acquire human state patterns, which are difficult to design in advance. As a result, we showed that state classification is possible using human skeletal information and density-based clustering.