Positioning of the mobile robot LiAS using natural landmarks and a 2D range finder

In this paper a Kalman filter based position estimation module with natural beacons is presented. The positioning module for the mobile robot LiAS has a flexible and generic structure to allow the contribution of several types of available external sensor data in order to reduce the position error. A special case is described in detail. The data of a 2D range finder is used to build geometrical primitives which are matched with the primitives of a known model. A matching primitive results in geometrical constraints which are fed to a Kalman filter resulting in a better position estimate. The parameters describing the geometrical primitives are provided with uncertainties which are used in the matching phase and in the Kalman filter. The basis of the position estimation is an enhanced dead-reckoning module which combines encoders with gyros. Further in this paper, an initial localisation algorithm is presented. The localisation algorithm uses a single range scan of the environment to estimate the robot position in a global coordinate system. Promising real-world experiments involving the positioning modules on the mobile robot LiAS are described