GPS 기반 옥외용 로봇의 인간 추종 기술 개발

Abstract

Recently, various trials have been made to extend the applicability of the robotics technology in service area. Human following robot has been the most interesting research topics in this area. The purpose of the human following robot is to follow the trajectory of human. The human tracking robot should always be aware of the position of human beings and should have the ability to travel autonomously and reach that position. Thus, localization is the crucial issue for the guide robot researches because autonomous navigation is strongly dependent on the localization performance. Another important issue for realization of the human following robot is how the robot always perceives the position of a human in an environment with many obstacles or in sloping terrain. Th conventional methods that use image or range sensor such as a camera or a ultrasonic sensor could not solve this problems. This thesis suggests an outdoor localization, a human position recognition method and a human trajectory tracking method based on GPS. For localization in outdoor environment, we develop an effective localization by using inexpensive GPS and odometry without using any INS system. Position and orientation information from DGPS and the digital compass are combined with encoder data from robot's wheels to estimate more accurate position of the robot using extended Kalman filter. The concept of "validation gate" will be introduced to ensure the reliability of the measured values for the error of DGPS. The thesis also proposes a method to autonomously follow the human trajectory that is generated by connecting the positions of humans. In order to show the feasibility and effectiveness of the methods, we also present experimental results of various situations using a real robot in a real world.