Development of an exploration system for a vision-guided mobile robot in an unknown indoor environment

• Main Authors: Jian-Hong Lai, 賴建宏
• Other Authors: Jong-Hann Jean
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/54737568870602844254

碩士 === 聖約翰科技大學 === 電機工程系碩士班 === 94 === In this thesis we present the development and implementation of an indoor exploration system for a mobile robot. According to the types of sensors utilized, we develop one exploration strategy by utilizing range-finding ultrasonic sen-sors and another exploration strategy by additionally incorporating vision sen-sors to navigate the mobile robot moving along the wall baselines. For the former strategy, we first use the ultrasonic sensors to measure dis-tances from the mobile robot to the wall in 26 equally-spaced directions around it. According to the gathered geometrical information of the environment, we apply the wall-following strategy to design a decision tree for determining the navigation strategy of the mobile robot. Then we design a steering controller based on the potential field method to drive the mobile robot moving along the wall baselines and keeping a certain distance from the wall. In addition, we de-sign an obstacle avoidance algorithm for the mobile robot to avoid collisions. With the aid of vision sensors, we propose another indoor exploration strategy. We first implement a visual tracking algorithm based on the random-ized Hough transform technique to detect and track the line parameters of wall baselines of the environment. Then by utilizing the feedback linearization tech-nique, we implemented a visual servo controller to successfully navigate the mobile robot following the wall-baseline and keeping a certain distance to the wall. Finally, we apply the indoor exploration system on a mobile robot to patrol the second floor of EE building of our campus for validating its performance, from which we can see the potential applications of the system as a security and surveillance robot.