| Optical navigation techniques are proposed as a solution to the problem of navigating in aplanetary system with features of real-time, autonomy and accuracy. Based on asteroid rendezvous,this paper focus on studying on autonomous optical navigation technology for deep space probe,details are as follows:Firstly, the method of simulation images generation is proposed to evaluate the suitability andeffectiveness of image processing algorithm presented this paper. Considering deep space celestialbodies optical image-forming principle and texture features of general celestial bodies such as flatsurface, crater and atmosphere, the simulation images are created. Then histogram analysis methodis used to make a comparison with real images.Secondly, image processing and information extraction algorithm is presented. The methodcombines image pre-processing technology and Canny edge detection algorithm for first edgesidentification, and the image gradient analysis is considered to eliminate the fake edge between thelight and dark areas. Least Squares Method, Levenberg-Marquardt Algorithm, and Improved DirectLeast Squares Estimation are separately applied to celestial body curve fitting, and the centroid ofgeometric outline is used for calculating the vector of LOS (Line-of-Sight).Finally, after a brief introduction of navigation camera imaging system, within selecting appropriatereference celestial bodies and navigation measurements, the navigation systems are built, whichincludes target asteroid LOS observation, target asteroid and the sun LOS observation, and targetasteroid LOS and subtend chord angle observation model. To evaluate the ability of measurements insystem state estimating, numerical analysis is used for observability and observability degree analysis.Complete navigation filters design based on Extend Kalman Filter (EKF) algorithm and stateestimations, which effectively improves the optical navigation accuracy and meets the requirements inthe deep space exploration mission. |
PDF Full Text Request