시뮬레이션을 이용한 이동 로봇의 충돌회피 알고리즘 성능 분석

Abstract

In general robotics researches, one of their final causes is to develop robots which can safely execute an assigned task even in dangerous and crowded environments. A service robot assisting humans must be able to promptly respond to unexpected changes in surrounding environments and freely move even in various external environments. A mobile robot must have an ability of autonomous navigation to fulfill its assigned role. For its autonomous navigation, the following essential technologies are required. First, sensor configuration technology for surrounding environment recognition and hardware configuration technology of a mobile robot system for autonomous navigation are needed. Second, path planning technology to select an efficient migration path and implement navigation in a task space is required to successfully complete an assigned task. Third, localization technology to acquire information from sensors mounted on a robot or in a task environment and analyze its current location according to given map data is needed. Last, collision avoidance technology to detect a surrounding obstacle and avoid collision is required to allow a robot to safely move from a start location to a goal location. According as the mobile robot has mobility due to its peculiarity, pre-verification of algorithm on motions of robots is important to secure stability. In the course of a task for verification of their motions, unexpected problems may occur all the time. Recently, various robot simulators has been used to verify algorithm, prior to application of a developed robot algorithm to a real robot, to previously prevent these problems. This paper applies essential technologies for autonomous navigation of the mobile robot, such as artificial potential field algorithm that was developed for collision avoidance and has been recently and widely used, elastic force algorithm and algorithm applied with the concept of a virtual sensor, to the same environment as that of the real robot, and their characteristics and performances are analyzed through simulation. For experiment, it also uses robot simulators including mobility of a differential drive mobile robot. According to development of a program to numerically analyze the distance the robot moves from a start location to a goal location, its total migration time, its distance from obstacle in case of collision avoidance, its smoothness and so on, by real collision avoidance algorithms, in case of their application to various simulators, to verify these algorithms, differently from the existing simulation programs, this paper analyzes each characteristic of the collision avoidance algorithms by using the above algorithm analysis program. To analyze characteristics and performances of the collision avoidance algorithms, it conducts numerical comparison of the distance the robot moves from a staring point to a goal point, the migration time, the distance from obstacle and the smoothness of the robot.