Self-localization of a mobile robot using fast normalized cross correlation

A self-localization concept for a mobile robot is presented, which is based on angle measurements to both known and unknown landmarks. The main contributions of the paper are the following: a fast normalized cross correlation algorithm (NCC) that uses a sum expansion of the template function and tables containing the integral over the image function (running sum) to detect the landmarks; and a linear solution for the relative position and orientation update of the robot using angle measurements to unknown landmarks. Furthermore, we apply a new kind of estimator for optimal sensor data fusion in the presence of both stochastic and deterministic errors for self-localization of the robot. Experiments demonstrate the feasibility of our approach.