Autonomous Navigation Method For Planetary Descent Stage Based On Image Information

The characteristics of planetary exploration is long distance and requiring high precision,optical navigation based on image information has advantages of strong autonomy and high precision,and will be one of the main navigation in the future deep space exploration.According to optical navigation of descending stage,this dissertation studies template matching of features of images,tracking of multi-frame feature points and position and velocity estimation of detector.The main contents of this dissertation are as follows.Firstly,this paper establishes dynamic model of descending stage,and designs the scheme of autonomous navigation based on image information.Considering the characteristics of detector descending stage,the dynamic model is simplified.Perspective projection based on optical observation is researched and the cheme of automous navigation for detector is proposed according to the characteristics of features navigation of planetary surface.Secondly,this paper presents the determination method for position of the detector based on template matching.Considering craters on the planetary as the navigation signs,images acquired when descending with previous obtained model of the planetary based on template matching are successfully matched,and according to this,mapping relations between navigation feature images and three-demensional positions are established,then the position and attitude in the coordinate system fixed-connected plnetary of the detector are solved using iterative least-squares method.Thirdly,this paper proposes an algorithm for motion estimation of the detector based on multi-frame image sequences.Considering the characteristic of detector descending,a method about corners detection and matching on the inter image is researched,which successfully matches the corners among multi-frame images.Then a scheme about motion estimation based on multi-frame image sequences is established,which determines the relative position by designing the match between every frame and the initial frame and the relative velocity by designing the match between neighboring frames.Finally,this paper presents a method of data fusion between positon determination and motion estimation.Acoording to the scheme about autonomous navigation based on image information,fusion filter structures and constant kalman filter are designed,which fuses the position determination and motion estimation,and perform a comprehensive mathematical simulation.