An Improved Indoor Localization Method for Mobile Robot Based on WiFi Fingerprint and AMCL

This paper proposes an improved indoor localization method based on WiFi fingerprint and adaptive Monte Carlo localization (AMCL). Its localization process consists of two stages, coarse localization and fine localization. The coarse localization of WiFi fingerprint is convenient and fast, but not accurate enough. However, the fine localization of AMCL is computationally complex, but relatively accurate. Therefore, the integration of WiFi fingerprint and AMCL for localization can be more accurate and efficient. In the coarse localization, the fingerprint method based on KNN is utilized to locate the position of robot and send the estimated position and estimated error to the ROS robot. Then, in the fine localization, the localization result estimated by WiFi fingerprint is regarded as the initial state of AMCL which can locate the exact position of robot based on grid map. In addition, in order to prevent the error caused by the slip of drive wheel and the cumulative error of the odometer, this paper fuses the information of odometer and inertial measurement unit (IMU) by EKF to improve the proposed distribution. The experimental results show that the proposed method can shorten the iterative time and improve the global localization accuracy of robot.