Research On Visual Measurement Technology Based On Motor Algebra For Spacecraft Attitude And Position Determination

With the rapid development of aerospace science and technology, relative position and attitude determination of target spacecrafts occupies a more and more important position in many kinds of space missions, such as rendezvous and docking(RVD), on-orbit maintenance and so on. Visual measurement methods are widely used in the close approximation stage of two spacecrafts, which have the advantages of small volume, low cost, low power consumption. In this paper, visual measurement technology of position and attitude determination is researched in order to find some more practical algorithms with higher concise degree, better real-time performance and higher reliability.Firstly, traditional description tools for rigid body position and attitude are introduced, including rotation and displacement matrix, Euler angles and quaternions. Then a new mathematical tool known as motor algebra is presented, which can uniformly describe the relative position and attitude of rigid body. Then motion models of points and lines descripted by motor algebra are deduced. The original three-dimensional nonlinear spiral motions of rigid body now can be expressed as a more compact linear form. Besides, principles of monocular and binocular visual measurements are studied, on the basis of which camera imaging model and distortion correction model are established.Since cooperative and non-cooperative target spacecrafts have different characteristics, this paper accordingly design different visual measurement methods for their position and attitude determination. Usually there are several pre-set light points on cooperative spacecrafts with known size and location. We use binocular vision to recognize these points and then calculate position and attitude through their pixel coordinates. As for non-cooperative target spacecraft, a new measurement scheme based on multi-sensor is proposed, using monocular camera to identify one rectangular plane on spacecraft and calculate its attitude, and using laser rangefinder to obtain its relative position. This paper also deduces two new position and attitude calculating algorithms using motor algebra, which require less calculating parameters and possesses high concise degree and better real-timing. Simulation results show that their precision can satisfy the measurement requirements.At last, this paper builds a ground platform using Microsoft Kinect for position and attitude determination demonstration, and the algorithms proposed before are adequately validated.