Human Following Control for Mobile Robot by Estimating Human Posture with Weighted Least-square Method and Virtual Connecting Model

Abstract

This study describes the human position/posture estimation method and follow-up modeling for human-following robots. For the posture estimation of a person, we propose a posture estimation method that applies the weighted least squares method to the contour points around the chest of the person measured by the laser range sensor. Regarding the modeling of the tracking system, we propose a dynamic connection model in which the robot follows the leader’s back as the target coordinates by connecting the person and the robot with a virtual rod.