Research On Target Recognition,Positioning And Guidance System Of Six-Axis Robot Based On Machine Vision

In the new era of China's development,the Made in China 2025 plan is the first action program of the Chinese government to implement the strategy of manufacturing power.One of its ten key projects is robot technology.The machine vision control system will become the "brain" of today's high-end manufacturing equipment.Based on machine vision technology,this paper designs a visual recognition positioning and guidance function system suitable for a variety of industrial robots.It focuses on the edge-preserving gray-scale smoothing filter technology,the establishment of feature descriptors and the rapid calibration of the hand-eye,and uses the laboratory six-axis robot platform to conduct system functional tests through specific experiments.First of all,the necessary components of the general vision application system are researched.In combination with the needs of this subject,the basic design of the system is structured to determine the development language and software development platform.Next,in order to ensure the effect of image feature extraction,first design and implement the pre-processing process.Based on the comparison of multiple filtering methods through specific experiments,a multi-gray-level filter based on image segmentation is proposed Degree-level smoothing filtering,and the use of bilateral thresholds for Canny edge extraction.Secondly,in order to complete the feature extraction,the SIFT feature descriptor is combined with the FLANN nearest neighbor search library for fast template matching.Recognition experiments were conducted under different interference conditions,and performance verification was completed through comparison of experimental matching results;at the last part of the visual information processing module,the smallest circumscribed rectangle was used to determine the position and posture for the corresponding actuator.Then the imaging mechanism of the industrial camera is researched,and the mathematical modeling of each coordinate in the camera is carried out.In order to ensure the accuracy of the system,the possible distortion of the lens is also taken into consideration,and the camera used in the subject is subjected to distortion correction and internal parameter calibration under the consideration of lens distortion.After the hand-eye model is determined,a quick calibration model of the system's hand-eye relationship is studied,and the accuracy verification and error analysis of the calibration results are performed.Finally,each functional component completed above is used to transmit coordinate values using the communication protocol related to the robot controller and the vision module.Through specific identification and guidance experiments to verify the performance of thesystem,the accuracy analysis is then carried out,the defects in the existing design are found in time,and the system design is continuously improved.