Principy řízení mobilních servisních robotů

Abstract

Laser range finders are frequently and successfully being used for localization of mobile robots. In this thesis we study localization methods based on scan matching and simultaneous localization and mapping (SLAM). By way of introduction the existing scan matching methods are introduced and divided into several groups. Especially point to point-based methods, feature to feature-based methods, classical methods and heuristic methods are described. The kernel of this thesis is the design of a new point to point-based method, which is based on correlation principle. The basic element of the new method is the suitable representation and calculation of correlation coefficient for three dimensional crosscorrelation. The optimizations, which decrease computation cost, are also included. Finally the experimental results and comparison with existing methods are proposed. The experiments have to particularly prove the dynamic objects robustness and the environment properties robustness and also the computation cost.