An algorithm for localization and positioning using linear combinations of model views

This paper proposes an algorithm for localization and positioning of robot in an indoor environment. The method is based on representing the scene as a set of 2D views and predicting the appearances of novel views by linear combinations of the model views. The method accurately approximates the appearance of scenes under weak perspective projection. An analysis of this projection as well as experimental results demonstrate that in many cases this approximation is sufficient to accurately describe the scene. The method has several advantages over other approaches: 1) it uses relatively rich representations; 2) the representations are 2D rather than 3D; and 3) the localization can be done from only a single 2D view