Mobile Robot Localization Based On Low-Cost LTE And Odometry In GPS-Denied Outdoor Environment

Abstract

GPS localization has always been the go-to method for localizing mobile robots in outdoor environments. However, in GPS denied environments such as urban canyons, LTE becomes a better alternative. LTE localization exploits existing infrastructures and transmitted signals to provide an estimated position of the robot. As a low-cost solution, it benefits robots under the constraint of cost, size and weight. This paper proposes a particle filter based localization method by using only LTE and wheel odometry for GPS-denied outdoor environments. We used the fingerprinting method by obtaining LTE Cell ID, mean RSS and GPS location and associating these data to the grids in an initialized map. We resolved the position of the robot using recursive Bayesian estimation and particle filter for its implementation. In our experiment, we used a mobile robot and five smartphones to obtain the wheel odometry and LTE data respectively while travelling along a particular route in an outdoor environment. The method was able to obtain accurate localization results with RMSE of 13.07m. We further evaluated the parameters of the method effects on the localization accuracy achieved.