Localization And Navigation System Design For Mobile Robot Based On Artificial Landmark

Localization and navigation technologies are the cores in the research area of autonomous mobile robot. Compared to the other navigation ways, visual navigation has the advantages of lager capacity of information and lower cost. It enables path setting more easily and has more flexible and reliable system. Visual navigation has become a popular research topic in the field of mobile robot navigation and localization. The ability of self-navigation and self-localization relies on a reliable way to identify the landmarks. Artificial landmarks have been widely used for their shapes, dimensions and colors could be preset according to the demands. In order to make mobile robot recognize landmark information precisely and quickly, as well as promote the ability of self-navigation and self-localization, a new-type artificial landmark and its corresponding recognition algorithm are designed in this paper. Then the posture and location information of mobile robot can be obtained, so that self-localization and self-navigation of mobile robot can be realized.According to the summary and the actual situation of the project, a visual servo navigation system is established, which includes mobile robot platform, visual sensors, artificial landmarks, personal computer and motion controller, etc. The landmark is recognized by visual system and the identification results are feed back to motion control system, then adaptive adjustment of mobile robot is made accordingly. The whole system has high positioning accuracy, fast response speed and good real-time performance.A new-type artificial landmark which is easier to be detected and identified is designed by analyzing the characteristic advantages of QR code. Then the method of self-navigation and self-localization is proposed based on the new-type landmark.On the basis of DirectShow and OpenCv function library, image processing is conducted on Visual C++ 6.0 software platform, including noise-reduction processing, peripheral logos detecting, binarization processing, feature recognition and information extraction,etc.In order to achieve better results in noise-reduction processing, an improved hybrid median-mean filter algorithm is proposed, which can well resolve the confliction between noise suppression and image details protecting. Before binarization processing, peripheral logo of landmarks is recognized and the extracted area is enlarged to enhance the image features, which is helpful to separate landmark image from background image. The resolution of landmark image is enhanced and the amount of data is reduced, which establishes a solid foundation for binarization processing. Feature identification algorithm is designed with reference to landmark features. And perspective distortion correction is performed on geometrically distorted images, so that regular landmark images can be acquired and the accuracy of information identification can be increased as well.Finally, static and dynamic experiments are performed on the mobile robot platform, which verified the reliability and effectiveness of the visual navigation system proposed in this paper.