Vision-based Navigation of Mobile Robots

• Main Authors: Jen-hung Chen, 陳貞宏
• Other Authors: Jane Yung-jen Hsu
• Format: Others
• Language: zh-TW
• Published: 1994
• Online Access: http://ndltd.ncl.edu.tw/handle/10029259420838587909

碩士 === 國立臺灣大學 === 資訊工程研究所 === 82 === The objective of this research is to explore the utility of vision processing for mobile robots with limited world knowledge.In particular, the ability to follow signs allows an agent to accomplish his assigned tasks in a structured environment. An efficient algorithm was developed to recognize a variety of signs in real time, about in one and half seconds. The technique was tested on a Nomad 200 mobile robot to carry out sign-based navigation in our lab building. In addition, the robot navigation based on vision-based recognition and path planning is applied in simulation world. The mobile robot is expected to be able to navigate much like a human in buildings. The mobile robot plans to go to some place and firstly it could generate a path to its goal based on a map of the environment. On the midway, the robot may meet accidents which prevent it from moving ahead. At this moment, the robot should successfully and efficiently recognize the signs posted on walls and follow the instructions of signs to detour these accidents. If the exit of instructions is on the original path, the robot tracks the path toward the goal position. Otherwise, it has to generate a new path from the current position to the goal and move toward the goal along this new path. The results show that the sign-following agent is useful and the sign- following with planning agent is intelligent and flexible.