Design And Implementation Of Storage AGV Navigation And Location System Based On Multisensor Fusion

In recent years,with the rapid development of information technology,automation and artificial intelligence,the modern logistics industry presents the trend of digitalization and intellectualization.Warehousing is an indispensable part of modern logistics.Automatic Guided Transport Vehicle(AGV)is the main equipment for cargo handling in intelligent warehouse.It is safe,efficient and labor-saving.It is an indispensable part of modern logistics and industrial automationThis thesis takes the intelligent logistics warehouse as the research background.the AGV cargo transportation link as the research focus,the other links as the auxiliary means,the Linux system,OpenCV(Open Source Computer Vision Library)and ROS(Open Source Robot Operating System)as the development platform.It combined with the practical application scenario,studies the design and implementation method of the warehouse AGV navigation positioning system with multi-sensors.High-precision sensors are expensive and inexpensive sensors can not meet the accuracy requirements.As the existing AGV single navigation strategy can not adapt to complex warehouse environment,being low efficiency and poor stability,A multi-sensor navigation and positioning method is used to divide the logistics warehouse into shelf area,connecting platform area.narrow corridor and wide channel in this thesis The shelf area and the connecting platform area are positioned by two-dimensional code.the narrow corridor is positioned by the combination of color band and two-dimensional code navigation,and the wide channel is positioned by laser navigation.On the basis of summarizing the advantages and disadvantages of single navigation mode,the thesis outlines the method of sub-regional navigation and its advantages over single navigation.Firstly,the navigation strategy combining two-dimensional codes and ribbons is studied and implemented,and the image feature processing method and the line-following correction algorithm are clarified.Visual image acquisition is susceptible to the influence of illumination changes,and the extraction of color band information is susceptible to the interference of ground stains and debris.Aiming at the above problems,an automatic threshold and region growing algorithm is proposed to eliminate the interference,reduce the impact of environmental changes on the system and enhance the robustness of the system.Secondly.the general method of laser navigation and the method of fusion with reflector are introduced.Finally,two precise connection control strategies,zigzag approximation method and fork approximation method for two-wheel differential AGV are studied.This thesis takes two-wheel differential AGV as the research obj ect,builds several indoor test environments and carries out field verification.The experimental results show that the two-dimensional code and ribbon tracking strategy has the characteristics of high accuracy,good stability and suitability for complex scenes.The laser navigation precision is improved obviously after the fusion of reflective panels,and the connection control based on the two-dimensional code also meets the predetermined requirements.