Research On Relative Motion Measurement Of Rocket Separation Based On Vision

The measurement methods based on visual has been researched extensively and applicated in space target detection,rendezvous and docking.However,the research and application of the relative position measurement of the separation of the rocket are less.Based on the previous research,this paper puts forward a method for measuring the relative position of the rocket based on computer vision.The pose measurement of moving objects based on visual measurement is to acquire images containing moving targets which has six degrees of freedom(Three position components and three angular components)by monocular or multi camera whose parameters has been calibrated.By establishing the three-dimensional coordinate system on the moving object or the moving target in a coordinate system with six degrees of freedom is known,the pose parameters(3D translation and 3D rotation)of the target coordinate system defined on the target body relative to the camera coordinate system can be calculated.The coordinate relation between the camera coordinate system and the reference coordinate system defined on the carrier of the camera group can be obtained by the calibration method.Finally,the relative position and orientation of the coordinate system of the moving target relative to the reference system can be obtained with the measuring camera for transmission.In order to measure the relative pose of space moving object based on computer vision,in this paper,the method and algorithm of rocket booster separation,fairing separation,interstage separation and relative position and attitude measurement are studied by taking four separate stages of rocket firing.The semi physical simulation experiment of ground has been established and carry out the simulation and semi physical simulation experiment.The purpose of the method and algorithm studied in this paper: According to the image information of the target and other prior information,the relative position and attitude of the moving object relative to the camera is solved.The influence of different estimation algorithms and the number of cooperative markers on the accuracy of the measurement results when using the control point method to estimate the pose,and how to measure the pose of the target according to the morphological features of the object,such as the circular feature without point constraint when the cooperative landmarks on the moving object are invisible.This paper mainly includes the following aspects.1)Introducing the model of camera imaging which is the center of the perspective projection model and the coordinate system involved in the measurement process and the transformation relationship between these coordinate systems.The camera calibration technology is studied,including camera intrinsic parameters,aberration coefficients,external parameters,and camera calibration method.2)The basic principle of position and orientation measurement is studied.The relative pose measurement method of the rocket's four separate stages is introduced.Taking satellite rocket separation as an example,the monocular and binocular pose measurement methods based on control points are analyzed.And the method of position and orientation measurement based on feature circle is introduced for the separation between stages3)Designing simulation and semi physical simulation experiment of ground.Experimental results show that the relative position error of monocular pose estimation based on control points is range from 0 cm to 1 cm within the range of 0 m to 6 m and the relative attitude is range from 0°to 3°within the range of 0°to 21°.The intersection error of binocular pose estimation based on control points is range from 0 cm to 2 cm within the range of 0 m to 10 m,and range from 0 cm to 1 cm within the range of 0 m to 6 m.The relative position error of monocular pose estimation method based on feature circle is range from 0 cm to 1 cm within the range of 0 m to 4 m,and the relative attitude of range from 0°to 2°within the range of 0°to 25°The method and algorithm of this research compared with the traditional method has more advantages such as low cost and high accuracy,and can be changed according to different application scenarios.It is also suitable for other applications such as target's position and pose measurement based on visual and 3D reconstruction,motion analysis and so on.