Outdoor navigation of a mobile robot between buildings based on DGPS and odometry data fusion

The authors aim at map based outdoor navigation of a mobile robot. In navigation, robot position is fundamentally obtained by odometry. However, the position is misaligned as the robot moves because odometry has cumulative error. DGPS measurement data may cancel its position error. The framework of EKF is used for the modification and the fusion between odometry and DGPS measurement data. The DGPS measurement data, however, could have large error because of multipath near buildings. In this paper, the authors propose a method which eliminates erroneous DGPS measurement data when odometry robot position is fused, and confirm the validity of this approach.