Research On Crater Matching Based Navigation Method For Lunar Precise Landing

The precision of navigation could be boosted largely when craters are used as navigational landmarks in lunar soft landing since their distinct features. This dissertation focuses on the crater matching based navigation method of lunar precise landing. The contents mainly consist of crater detection, crater matching and crater matching based navigation.Aiming at the navigational application, we present an edge information based crater detection algorithm, which consists of edge detection, edge selection, edge pairing and ellipse fitting. For the problem of poor adaptability to imaging conditions, the algorithm is improved by introducing Multi-scale Adaptive Gaussian Filter and Adaptive Canny Operator based on image gradient magnitude histogram, which results in Adaptive Crater Detection Algorithm. Experiments of real lunar images show the improved detection algorithm is provided with adaptive ability to different illuminations and terrains of the crater images, which solves the problem of small tolerance to imaging conditions of former algorithms. For crater matching, a voting strategy based algorithm is studied by using Conic-pair Affine Invariants, and a correction algorithm is suggested for mismatching and false matching situations by using affine property of close curves'areas. The experimental results of multi lunar images show the matching ratio of the algorithm is higher than 85% with lower 5% mismatching.For crater matching based navigation, we present an optical measurements based relative navigation method, from which the position and attitude imformation of a lander can be determined by measuring the lines of sight of 4 or more than 4 mapped landmarks. Considering the respective limitations of optical navigation and inertial one in practical application, we suggest a Crater Matching/IMU based combinatorial navigation algorithm. This algorithm increases the updating rate of navigational outputs by integrating optical measurements with IMU using Extended Kalman Filter. Precise angular velocity could be acquired since the drift error of gyro could be corrected by optical measurements, which is available for real-time control. Results of simulation show the proposed navigation algorithm is provided with a high accuracy of meters-level position error, which meets the requirements of precise landing.