Implementation of Localization System using Learning Automata based Sensor Fusion in Unmanned Forklifts

摘要

Unmanned forklifts have great potential to enhance the productivity of material handling in dangerous applications because these forklifts can pick up and deliver loads without an operator or any fixed guide. There are, however, many technical difficulties involved in developing unmanned forklifts including localization, map building, sensor fusion, and control. Recently, the NAV200 positioning system has been used as a localization system, which is the most important component of unmanned forklifts. The NAV200 is a laser measurement system for indoor localization with high accuracy and high precision; however, it has some problems in that it may not operate well when it is required to move fast or has to change its direction at an instant. In order to solve these problems, this paper proposes a learning automata based sensor fusion algorithm with dead reckoning using the kinematics of the unmanned forklift and Kalman filter based prediction using the tendency of movement. To demonstrate the feasibility of the suggested sensor fusion algorithm, its performance is evaluated in computer simulations for various cases.