Landmark Localization System For Outdoor Mobile Robot

Autonomous mobile robot is an important field in research and application of robot. While its application area develops faster and faster, the robot is supposed to replace human being to work in the remote or adverse outdoor environment, such as electric substation and chemical plant which faces high temperature, tall radiate and strong interference. As the foundation of robot working in outdoor environment, localization becomes the first problem to solve.An outdoor localization system for mobile robot based on landmark recognition is presented. The system is robust under the changes of lighting conditions. The near infrared illuminator is used for reducing the interference of natural light and shadow. Omni-directional camera can capture images with a large view of environment. In the image processing, dynamic Otsu method and landmark tracking make the landmark's recognition run more accurate. It also improves the rate of system. The triangulation approach is then used for robot's localization and navigation. The real-world and real-robot experiments demonstrate the system's accuracy and robustness.So, the study in this dissertation provides some innovative approaches in outdoor localization subjects.First of all, this dissertation proposes a vision system with near infrared illuminator, Omni-directional camera and artificial coded landmarks. The near infrared illuminator can reduce the disturbance of natural light. Omni-directional camera is adopted to capture images with a wide range of landmark's information. So, the system can adapt the influence of natural light and noise in complex outdoor environment.Secondly, this dissertation designs a robust near infrared landmark recognition algorithm. While Omni-directional image is captured, it is expanded to cylindrical panorama which is more popular for representing the environment. Then, dynamic Otsu method is adopted to binarize the image, which enhances the feature of landmark. At last, recognition landmark abstracting and template matching accomplish the landmark recognition.Finally, the dissertation tests and verifies the property of system through multiple experiments. Experimental results show that the near infrared Omni-vision system can effectively depress the nature light and shadow in outdoor environment. And the near infrared landmark recognition algorithm can recognize the landmark accurately and quickly. After localization with triangulation approach, the system shows nice location accuracy and robustness.