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

Abstract

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