A stochastic approach of mobile robot navigation using customized RFID systems

Abstract

Operating a mobile robot using the signal strength of a Radio Frequency (RF) system and/or line-of-sight distances to other known points or RF stations is a challenging task. This problem has been traditionally solved by several approaches suggested in the literature. Among the most common shortcomings of those approaches are the use of excessive number of sensors or multiple reference RF stations for the robot to estimate its location in an indoor environment. The current manuscript outlines two different aspects of a mobile robot navigation problem in an indoor environment using Received Signal Strength (RSS) of a customized Radio Frequency IDentification (RFID) system. First, the robot's current location is estimated by a trilateration method where the localization problem is solved through a geometric approach based on Cayley-Menger determinants. The robot position is then better estimated by the application of conventional stochastic filters such as the Extended Kalman Filter (EKF) and the Unscented Kalman Filter (UKF). Second, the problem is explored by a set of points on the ground defining a desired path along which a mobile robot is supposed to navigate. The proposed robot navigation system is validated through a number of computer simulation for testbeds of various complexities.