Research On Sensor Fusion Based Free-Path Steering For Automatic Guided Vehicle

Since the automated guided vehicle (AGV) was to be widely used in the fields of industry such asmanufacturing, the related technology also has made greatly development. As a wheeled mobile robot,AGV its system could be divided into five modules by the definition in Mechatronics, includingmechanical body, power-driven, implementing agencies, sensing and also control and informationprocessing. Depending on the difference of sensors equipped, there was Magnetic Tape Guidance,electromagnetic guided navigation, Vision Guidance, Laser Navigation and other means of guidancemethod. For the low cost, high precision, and high level of flexibility and extensibility, there was alarge growth prospects for Visual Guidance AGV.Based on the analysis of differential-driven AGV model and UMBmark calibration of the actualsystem, the main topic in this article was about using information fusion methods to achieve theglobal localization capability for the AGV system, and making it off guidance steering within afree-path.Different from traditional Guidance, on the basis of the vision guided AGV, this article focused onthe integration with the encoder feedback and RF tag information, and the model for AGV which dothe information fusion within multiple data sources was established. Kalman filtering method wasused to optimize the estimation that the AGV’s deviation to the guide line through the combination ofthe encoder feedback and the bias of line which extracted from visual signal. Then based on RFinformation, the position was ensured that which integral of the feedback from encoder could beupdated in every segment of the path, achieving the capability of global localization.Secondly, on the presupposition of analyzing the advantages and disadvantages of two modes inAGV’s steering, the free-path steering was proposed. By introducing the S-curve acceleration anddeceleration control method which is widely used in CNC system, the path for off guidance steeringfor the AGV could be planned. And an AGV system model was built in Simulink environment underMATLAB software and the free-path steering method was been validated in simulation. Finally, thereal system was programed and tested to validate the possibility and reliability of the design.