A multi-sensor fusion method based on EKF on granary robot

Abstract

In this paper, a set of algorithms of granary robot is designed. The traditional outdoor localization technology is not suitable for indoor because of the shelter of the wall of the building. Traditional indoor localization technology has a poor effect because of the continuous change of terrain in barn. Compared with various indoor localization technologies, ultra wide band (UWB) localization technology has the advantages of strong penetration, low power consumption performance and high localization accuracy, but the problems of the system is a low updating frequency. Because of a low update frequency of UWB localization system, the paper presents a joint localization system which fuse Inertial-Measurement-Unit (IMU) inertial navigation system and the UWB localization system based on Extended Kalman Filter (EKF). It makes full use of the complementary features of the functions of the inertial localization system and the UWB localization system. The inertial localization system can continuously process the UWB localization information and it compensates the problem of a low refresh frequency of the UWB module, continuously optimizes the position and attitude in the dynamic process. The experiment shows that this joint localization method breaks the limitation of moving speed and improves the localization accuracy. The dynamic localization accuracy error of the system is about 10cm, and the calculation error of the attitude is within 0.3 degree. The localization result meets the needs of the granary robot.