Indoor navigation of mobile robots based on natural landmarks

To derive cost efficient solutions, navigation of mobile robots should take advantage of known facts of the environment, like maps of buildings. This paper discusses the use of range sensor measurements at characteristic locations to periodically correct the drift in odometry data, in order to derive the vehicle's position and orientation with sufficient accuracy. The main objective is to provide a low cost navigation system without the need for external reference marks. This paper outlines the approach, the sensor data processing algorithm and results of performance tests in industrial environments.