Research on Indoor AGV Fusion Localization Based on Adaptive Weight EKF Using Multi-sensor

Abstract To solve the problem of large accumulative errors wheel odometer and inertial measurement unit (IMU), which make pose calculation mistakes due to friction coefficient ground or bumps collisions, furthermore, affect path planning navigation automated guided vehicle (AGV), an extensible fusion localization method based on multiple sensors is proposed. Based traditional extended Kalman filter (EKF), internal positioning data used for prediction, while external correction. According states different sensors, adaptive weight applied eliminate influence system error provide continuous accurate data. The experimental results show that compared single-sensor, has increase in quantity, can reach accuracy within 5~8cm. It deal with non-line-of-sight (NLOS) landmark occlusion situations, navigate dim light dark environments. ensures through visual stability IMU. New are also possible be added into fusion.