摘要：The self-localization for a mobile robot is very important at indoor environments. In this paper, we propose a relative self-localization estimation algorithm based on relative locations and orientation changes of image features. We also analyze errors caused by a variety of factors to estimate the relative self-localization of a mobile robot and discuss a few techniques to remove them. The proposed relative self-localization algorithm is based on the facts that the global orientation and location of image features are not altered by changing of images. We show that the proposed algorithm is valuable through some simulation examples.
