| The research of outdoor mobile robots is an important part of robotics research,which includes two fields, structured road and unstructured road. Vision navigationunder structured road has become the most important part of the civilian purposeresearch,and many successful systems have been invented. There are also manyresearches about unstructured road, but this field is not mature on basically. Althoughthe technology of image detection and the capability of image comprehension canmake the mobile robots work in special environment, there's long way to go to realizethe really adaptive capability.The paper is concentrated on two parts. On the aspect of structured road, visualnavigation in multiple scenes, especially the vehicle detection method, is researched.The paper presents an new approach based on multiple characters: a primary area ofinteresting is found by detection of the shadows underneath the vehicle, then thecontour is found based on the vehicle symmetry measurement.The other part is the basic research on visual navigation for mobile robot inunstructured road. A method of road detection based on the theory of mathematicalmorphology is proposed, which can be applied not only on structured roads such asexpressway but also on unstructured roads that have no lane markings, may havedegraded edges, and may be partially obscured by shadows. Through analysing theroad characters, performs morphological opening operation on the image to separatethe interferential area such as trees and sky from the road suface, finaly use the rigionsegmentation algorithm to find road suface out. This method is influenced little byshade and illumination. |
PDF Full Text Request