확장 칼만 필터 기반 지능로봇의 옥외용 3차원 위치평가 방법에 관한 연구

Abstract

As the fourth industrial revolution has become an issue in recent years, there is a growing interest in intelligent robot technologies. The application fields of intelligent robots are diversified into various fields such as robots that work on behalf of human beings in disaster areas or accident sites, autonomous driving cars that do not require drivers. In these applications, the evaluation of robot's precise position is the most important part in order for the robot to accurately perform a given task. Such a position evaluation technique of a robot is called localization. The localization of the robot varies greatly depending on whether it is an indoor environment or an outdoor environment. In the indoor environment, it is possible to perform precise position evaluation by using an environmental map or various localization sensors. On the other hand, the outdoor localization is very different from the indoor localization because the outdoor environment is difficult to use the environmental map or install the localization sensor. Therefore, it is general to use GPS to evaluate the position of the robot in an outdoor environment. In addition, the localization of the indoor environment is sufficient for evaluating the two-dimensional position because the floor is flat, while the localization in the outdoor environment requires the three-dimensional position evaluation because there is a slope. Until now, a variety of localization techniques have been developed for 2 dimensional localization, but there has been little research on the 3 dimensional localization. 3D position estimation technology is essential for fields that require 3D position information such as caddy robots for golfers, military robots, and autonomous vehicles. Moreover, if only 2D localization is performed in an environment with inclination, the position error can be further increased by the altitude difference. Thus, in such an inclined environment, 3D localization can be more effective than 2D localization. This thesis proposes a 3D localization method for an outdoor mobile robot using only non-inertial sensors such as a DGPS, a digital compass and an inclination sensor. This method can evaluate the 3D position including the altitude information which is impossible in the existing 2D localization method. In this method, the 3D position of the robot is predicted using an encoder and an inclination sensor, and the predicted position is fused with the position information obtained from the DGPS and digital compass using extended Kalman filter to evaluate the 3D position of the robot. Experimental results showed that the proposed method can effectively evaluate the 3D position of the robot in a sloping environment. In addition, this method was found to be more effective than the conventional 2D localization method even in the evaluation of the plane position where altitude information is unnecessary. Also, it was confirmed that the developed method is stable even if the size of initial error and observation period change. Therefore, it is expected that the proposed method can be applied more effectively not only in fields requiring 3D localization but also in fields requiring 2D localization.