A novel approach to the mobile robot localization problem using tracking methods

This paper presents a novel approach to mobile robot localization. This approach is basically different from other methods developed so far, because the sensors are not mounted on the mobile robot. In fact, they are statically located in the environment. These sensors, which could be for instance SICK lasers embedded in the walls of a museum, constantly measure the distance to points on the surface of the mobile robot in a plane. The only further assumption we have to make is that the robot has a circular shape in this plane. This supposition is feasible, since most indoor service robots have a circular shape. We then use a simple tracking algorithm based on the Kalman filter and geometrical considerations to localize the robot with a Very high precision. In simulations it is shown, that our method is able to localize the robot with an error in the range of one centimeter.