The AGV Guidance Control And Obstacle Detection Based On Vision

With the improvement of automatic level,the implementation of AGV system is more and more widely and plays an increasingly important role.The successful implementation of AGV system can greatly reduce the cost of human resources and increase production efficiency.In the past few years,with the development of handware such as computer,sensor technology,etc,the vision-guided technology becomes a research focus.In this paper,the visual navigation and obstacle detection methods of AGV are studied,to transport the material in the actual assembly workshop as the scene,aiming at some practical problems such as non-uniform illumination of the assembly workshop,damage of the navigation marks and obstacles,etc.The main contents and achievements are listed as follows:1.Image processing and recognition of navigation belt.A new extraction algorithm of the center line of the navigation belt under complex conditions is proposed,aiming at the problems caused by non-uniform illumination,damage of the navigation marks,such as incompleteness and imprecision of the visual guidance belt center line,etc.Experimental results show that the algorithm can eliminate the influence of above factors and realize extraction of center line effectively.According to acceleration,deceleration,turning,parking and other practical functional requirements in the traffic scene,designed a series of navigation identifiers and proposed an image recognition algorithm based on improved Hu invariant moment,can lead AGV to identify the corresponding navigation identifier correctly and effectively.2.Navigation control method and obstacle detection.In the non-obstacle path,AGV should be fast and efficient and keep high precision.PID control effect is not ideal for nonlinear system.In the paper,the fuzzy PID control method is adopted.The softwaresimulation and experimental comparison of PID and fuzzy PID control are carried out.The results show that the fuzzy PID control effect is better.When there are obstacles,AGV requires effective identification and acquire the information of obstacles.An obstacle judgment algorithm based on binocular depth detection is proposed.At the same time an algorithm to obtain the information of obstacle orientation and speed is proposed.Experimental results show that the algorithm is able to identify obstacles effectively and obtain its parameter information.3.Experimental platform construction and algorithm verification.The experimental device of AGV is set up.Design and develop a system control software based on c# and EmguCV.Respectively,do experiments for centerline extraction algorithm,the navigation belt characters recognition algorithm,no obstacle path control algorithm and obstacle detection algorithm,results show that the proposed algorithms are effective.