Measurement Method Research Of Robot Repetitive Positioning Accuracy Based On Monocular Vision

The repeated positioning accuracy of the robot is the basis of the absolute positioning accuracy of the robot,and it is also the premise of the normal operation of the robot.Therefore,the research on the measurement method of the repeated positioning accuracy of the robot has been paid more and more attention.At present,the measurement of the repeatability of the robot is usually measured by a laser sensor or a large-scale high-precision measuring device.These devices are expensive,complicated in operation,and have strict requirements on the measurement environment.In this context,this paper proposes a method for measuring the repeatability of robots based on monocular vision.The purpose is to reduce the complexity and cost of the measurement and improve the measurement efficiency.The main research contents of the dissertation are as follows:(1)This paper compares and analyzes various image preprocessing methods,feature point localization methods,image feature extraction methods and camera imaging models,and combines the advantages and disadvantages of the above methods with the application scenarios to determine the adaptive median filtering and histogram equalization and adaptive threshold segmentation algorithm to preprocess images.This paper uses ellipse fitting method to locate feature points and uses Canny operator edge detection and Hough line detection method to extract image features.Select the central perspective projection model according to the application scenario of the measurement method.(2)Aiming at the practical application requirements of robot repeating positioning accuracy measurement,the overall scheme of the system is designed.The system includes hardware selection and software platform design and development.This paper compares and analyzes multiple camera calibration methods,and proposes an improved self-calibration method based on line detection based on the characteristics of measurement methods.(3)The target pose calculation algorithm is analyzed from the aspects of computation speed,calculation accuracy and accuracy.Combined with the characteristics of the end of robot motion,this paper proposes an improved target pose solving algorithm that achieves a balance between computational speed and computational accuracy.(4)Build a robotic repeat positioning accuracy measurement experimental platform,and offline calibrate the internal and external parameters of industrial cameras.Completed the robot repeat positioning accuracy measurement experiment,finally,an error analysis is performed on the entire measurement system.