Visual Navigation Algorithm Of Lander Based On Irregular Curve

In recent years,countries have launched landing and sampling return missions for different celestial bodies,which has drove the development of visual navigation algorithms based on feature matching.However,visual navigation algorithms based on traditional navigation landmarks have the disadvantages that they cannot be used for absolute navigation and cannot be used when there are few features.Irregular curve feature is a ubiquitous feature on the surface of celestial bodies.It can be used as a navigation landmark for visual navigation algorithms,and can be used for absolute navigation.Based on this,this paper studies the visual navigation algorithm of the lander based on irregular curves.The main research contents of the paper are as follows:First of all,according to the difference of celestial body surface curve characteristics,a crater feature extraction and matching algorithm based on feature descriptors and an irregular curve feature extraction and matching algorithm are studied.If there are many craters in the field of view of the navigation camera,by fitting the curve edges of the craters to circular arcs,an circular arc extraction algorithm based on curve growth is used to extract crater feature,construct a crater descriptor,and use Euclidean Distance and Nearest Neighbor Distance Ratio to match crater feature.If the number of crater features in the navigation camera's field of view is insufficient,use the Sobel operator and pixel stitching algorithm to extract the irregular curve features and generate an irregular curve descriptor,and then use the BBF query algorithm and the Nearest Neighbor Distance Ratio to match the irregular curve descriptor.The simulation results show that the above two algorithms can effectively extract and match the corresponding features,and have the invariance of scaling,illumination and rotation.Secondly,based on the matching results,the visual navigation algorithm of the lander based on irregular curves was studied.Based on the modeling of the lander navigation camera,two point-tangents on the three-dimensional irregular curves,the two-dimensional point-tangent on the image which matched with them,the navigationcamera model and the first-order curve differential relationship are used to establish equations.Simplify the equations to obtain a 16-degree equation in one variable,and solve it to obtain multiple sets of solutions.Finally,the unique solution is determined by the camera projection model,the position and attitude of the lander at the previous moment,and its motion status.Through simulation experiments,the influence of different noise intensities on the algorithm in this chapter is analyzed;through simulation of the landing process,the reliability of the algorithm's estimation results and navigation accuracy are verified.Finally,for the known three or more point-tangents on the three-dimensional irregular curves,as well as the two-dimensional point-tangent on the image which matched with them,the visual navigation algorithm of the lander based on overdetermined equations is studied.The equations are established by the matched known quantities,the navigation camera model and the first-order curve differential relationship,and simplified the equations to the form of overdetermined equations,and then the optimal solution is obtained by the least square method.The simulation results show that the three-axis position and attitude errors show a tendency to decrease with decreasing altitude.In the end,its three-axis position estimation error is within 0.5m,and its three-axis attitude estimation error is within 0.1 °.