Fuzzy-EKF for the mobile robot localization using ultrasonic satellite

Localization accuracy is a significant fundament for autonomous mobile robot navigation. In this paper, robot fuses the information from odometry and the ultrasonic satellite for localization. In order to improve the accuracy of localization, a Fuzzy-extended Kalman filter (Fuzzy-EKF) method is applied to avoid the robot using the large error data to update the position continuously. A weight scalar is designed to change the noise covariance by inputting the robot rotation angle, innovation and the measurement data variation into the fuzzy system. Therefore, the proportion of the system and measurement value changed, which decreases the robot state errors indirectly. The simulation results demonstrate the improved performance of the proposed Fuzzy-EKF method over the conventional EKF method.