New Data Fusion Techniques for High-Precision Localization of Civil-Engineering Equipment

This paper presents two localization fusion techniques for civil-engineering machines. The first one concerns 2D localization and the second one is intended to 6 dof position and attitude determination. The system uses an optical instrument called SIREM which provides azimuth and elevation angles of known landmarks. The approach is based on dead-reckoning techniques updated by the exteroceptive measurements of SIREM using the Kalman filter formalism. As the dynamics and accelerations of the vehicles we consider are small enough, the use of inclinometers is attractive since they directly provide attitude angles. In such a context, we propose a new 3D dead reckoning method which integrates data from two inclinometers and an odometer. A large section of the paper is devoted to real experiments. These are performed with an outdoor tire-type robot moving on a 30x40 square meters non planar lawn. The results show that the precision of our system, although difficult to assert with the current experimental setup, is in the order of centimeters.