Research And Implementation Of AGV Target Localization And Motion Control Based On Multi-Camera Global Vision

Automated Guided Vehicle(AGV)is an unmanned transport vehicle with automatic navigation function,which can complete the specified tasks according to the planned path.It is also the main equipment in the field of intelligent manufacturing and automatic transportation.It has the advantages of high flexibility and safety,and is widely used in medical,catering,military and other industries.At present,the development of visual guidance AGV has just started.Compared with traditional guidance AGV,visual guidance AGV has the advantages of high flexibility and strong adaptability to the environment.Compared with local visual guidance,global visual guidance can obtain richer environmental information and better perception ability,but due to the immature technology,there are still problems such as low map accuracy,poor real-time detection and poor robustness.Based on the above-mentioned problems,this paper makes a research,and the main contents are as follows:(1)In global map construction,the four feature points of a designated rectangular marker are identified to perform perspective transformation on captured images.The accuracy of feature point extraction is relatively high.Two small markers are placed within the rectangular marker for the purpose of identifying feature points and achieving image stitching to build a global map.Through experimental comparison,it has been found that this method enables fast matching and verifies the accuracy of the global map construction,with a maximum error of0.003%.The global map exhibits high precision.(2)In the AGV detection and positioning stage,a target detection and positioning algorithm based on color features was proposed.A color filtering model was established in the RGB space to recognize and locate targets by designing two different colored blocks on the AGV to calculate direction.This approach improved the AGV’s ability to withstand environmental interference,enhanced recognition robustness,and eliminated the influence of height on positioning accuracy.Through experimental verification,the AGV positioning accuracy was improved,with a maximum positioning error of 0.004%.(3)During the AGV path planning and tracking phase,a control platform based on C# was designed to control the AGV through a map stitching module,an AGV target recognition module,and a path generation and control module.To address the problem of deviation between planned and actual paths,a control method based on correcting deviation was proposed to adjust deviation and improve the agreement between planned and actual paths.Experimental analysis was conducted to compare planned and actual paths,and it was found that the proposed method improved their agreement,demonstrating its feasibility and reliability.The maximum tracking error of the path was 0.008%.