Comparison of Localization Algorithms for AGVs in Industrial Environments

Abstract

Estimating the position and orientation of Automated Guided Vehicles (AGVs) in industrial environments is an important and difficult task. The main problems of this task are the changing industrial environments and the requirements on the positioning accuracy for docking applications. In current AGVs, special sensors for localization, which can guarantee accurate positioning with a high repeatability, are used. The drawbacks are their prices, the requirements concerning the construction of the AGV and the additional required safety sensors. In this paper three frequently discussed algorithms for position tracking are compared: the Extended Kalman Filter, the Unscented Kalman Filter and the Monte Carlo Particle Filter. To reduce costs of an AGV, distance measurements of the required safety laser range finder (LRF) are used. The positioning accuracy and the time behavior of the algorithms are analyzed and the movement and measurement functions of the used omnidirectional AGV are presented.