Comparison of mobile platform localisation using odometry and visual information

The purpose of this thesis was to compare robot odometry with localization using visual information. In this thesis we used a robot iRobot Roomba series 531. Since this is a commercial robot, the price of such robots is relatively low, therefore odometry of such robots is very inaccurate. Localization with visual information was realised such, that we fixated color camera on top of a robot, which was directed toward the ceiling. On the ceiling, we attached four markers, through which robot movement was calculated, according to the captured image of ceiling markers in the initial robot position. The system was evaluated so the robot was given a pre-known path on which it moves, through which the error of robot odometry has been found. On the same path accuracy of visual information was obtained. At the end we compared errors of visual information and robot odometry