期刊名称:International Journal of Advanced Robotic Systems
印刷版ISSN:1729-8806
电子版ISSN:1729-8814
出版年度:2017
卷号:14
期号:1
DOI:10.1177/1729881417693489
语种:English
出版社:SAGE Publications
摘要:This article presents a self-localization scheme for indoor mobile robot navigation based on reliable design and recognition of artificial visual landmarks. Each landmark is patterned with a set of concentric circular rings in black and white, which reliably encodes the landmark’s identity under environmental illumination. A mobile robot in navigation uses an onboard camera to capture landmarks in the environment. The landmarks in an image are detected and identified using a bilayer recognition algorithm: A global recognition process initially extracts candidate landmark regions across the whole image and tries to identify enough landmarks; if necessary, a local recognition process locally enhances those unidentified regions of interest influenced by illumination and incompleteness and reidentifies them. The recognized landmarks are used to estimate the position and orientation of the onboard camera in the environment, based on the geometric relationship between the image and environmental frames. The experiments carried out in a real indoor environment show high robustness of the proposed landmark design and recognition scheme to the illumination condition, which leads to reliable and accurate mobile robot localization.