| AGV(Automated Guided Vehicle),also known as an autonomous path-finding intelligent vehicle,is mainly used in the Internet of Things scenario,and equipped with boot equipment,which can run independently to work stations to meet production needs.In essence,this business model is conducive to more flexible and intelligent production.In addition,AGV can also improve the registration rate of cargo transportation,increase safety,guide equipment to make adjustments according to demand,which is conducive to promotion.For the entire intelligent factory,it can greatly reduce human resources to save costs and improve production efficiency.The advancement of science and technology and the improvement of people's demand make the technology research of AGV develop rapidly,and the intelligent navigation of AGV is one of them.Nowadays,the main navigation modes of AGV include lidar,inertial guidance,tape navigation and so on.There are still many challenges to be faced:1.The high cost of laser navigation will limit its application.2.The cumulative error of inertial guidance will occur,and further research on correction technology is needed,otherwise it will hinder the long-term stable operation of equipment.3.Tape navigation is not easy to adjust the AGV production line and has poor flexibility.This paper mainly studies AGV's path recognition and tracking control in complex environments such as uneven illumination,ground cracks,slight path breakage and so on.Compared with the traditional AGV navigation mode,the use of color bands as guidance can greatly reduce the cost of navigation;secondly,the path modification is more convenient and more flexible;with the help of computer vision technology to accurately identify the path information to achieve AGV autonomous navigation,with a higher degree of intelligence.So the current visual navigation technology is hopeful to promote the new process of AGV intelligent navigation in the field of unmanned factory.The experimental results also prove the effectiveness of the AGV visual navigation studied in this paper.The main innovations and contributions of this paper are summarized as follows:1.In view of the interference information in complex scenes,which seriously affects the detection efficiency and robustness of path recognition,a geometric model is established according to the installation location of camera sensors,and a dynamic region of interest(DROI)is designed based on the transformation of spatial coordinate system and the cumulative probability Hough Transform(PPHT).The dynamic analysis window is built adaptively according to the size of AGV,which improves the computational efficiency and robustness of the algorithm to a certain extent.2.Aiming at the detection interference caused by uneven illumination,a method of edge detection based on YCr Cb is proposed,which combines OTSU algorithm to segment the marking band.In order to remove the noise better,the image is further sampled and the edge is detected by morphological operation.Finally,the noise can be effectively filtered and the edge of the identification band can be accurately extracted.This method provides the necessary basis for AGV path tracking control.3.Aiming at the problem of ground cracks and damage of marking band,an optimal model is established based on dynamic programming and PPHT to accurately identify path information.Finally,based on Hough Circle to locate bend marking,a decision-making control model is established to track AGV and realize intelligent navigation. |
PDF Full Text Request