Research And Technology Of Space Target Postioning According To Feature Points Identification

As a high-precision non-contact measurement method,visual pose measurement technology has a wide range of applications in spacecraft rendezvous and docking,onorbit maintenance,and space debris cleaning.Pose measurement technology based on monocular vision has the advantages of simple structure,low cost,and large measurement field of view,which fulfils the needs of on-orbit position and psoe measurement.In order to guarantee the actual position and posture state of the space development mechanism to reach the ideal situation,the visual measurement result was used to guide the control amount of the positioning system,so as to correct the measured object to reach the ideal position and posture state.The geometric constraint information provided by the point feature can simplify the measurement problem to the PnP(Perspective-n-Point)problem.The feature information of the measured object can be enhanced through the cooperative target which ensures the accuracy and stability of the position and pose measurement result.In recent years,research on pose measurement systems has focused on improving the accuracy and efficiency of the PnP algorithm,while ignoring the effect of increasing the number of feature points on the measurement accuracy.In this paper,a complete position and pose measurement model was established.By studying the influence of the feature point layout and cooperative target structure parameter on the measurement accuracy,a multi-point cooperative target was designed to quantitatively improve the measuring accuracy,and the functional analytical formula for the measurement accuracy of the n-point target was obtained.The measurement results showed that the reprojection measuring error of the designed 16-point ring-array target was reduced about 50.3% compared with the 4-point target of the same size.Also the reprojection measuring error of the designed three-dimensional 20-point cooperative target was further reduced about 39.2%.On this work,a monocular vision pose measurement system was built,and a complete experiment of camera internal parameter calibration,distortion correction,feature point recognition and matching was carried out efficiently.The data directly obtained by vision measurement can reflect the postion and pose state of the measured object relative to the camera.In practical applications,the original measurement data should be converted to the measured object space to provide guidance for the control amount of the positioning system.In order to solve the exsiting problems of traditional camera external parameter calibration such as cumbersome operation,additional equipment required,and not conducive to on-orbit implementation,a data processing method using the motion sample data space to construct the coordinate system conversion relationship was proposed here.According to the compensation equation constructed in the sampling space,the visual measurement data of six degrees of freedom could be converted to the controller coordinate system,and the system errors caused by installation and alignment could be corrected at the same time,thereby achieving the highly accurate space targeting.Experimental results showed that based on the above established data processing methods,the lateral position measurement errors of the single-camera measurement system at a measurement distance of 1.7 m were less than 0.02 mm,and the attitude measurement errors were less than 0.006°,which verified the effectiveness of the measurement system.