Research On Terrain Identification And Mechanical Parameters Estimation Of Lunar Soil

Unknown lunar environment requires that a lunar rover must have high ability of autonomous control, for which the determination of terrain tranversability plays an important role. In order to the determination of lunar terrain non-geometry traversability for navigation and control of lunar rovers, this dissertation focuses on two problems: terrain identification and terrain mechanical parameters estimation. For terrain identification, a new method of terrain classification based on wavelet multi-resolution analysis and maximum likelihood criterion is presented for identifying terrain non-geometry obstacles. Firstly, the filters are constructed based on Daubechies wavelet. Then Mallat wavelet algorithm is applied for multi-resolution decomposition of a terrain image, and the centroid of area of the high frequency energy distribution of the decomposed image is taken as the texture feature. Finally, non-geometry obstacles can be identified based on maximum likelihood criterion. Three group experiments demonstrate the robustness of the texture feature and the correctness of the algorithm.For terrain mechanical parameters estimation, a new method based on on-board sensors is presented. First, the Janosi shear module is simplified, and the force/torque balance equations of driven wheel are derived based on modules. Then, according to the properties of lunar soil, the mechanical parameters are determined. Finally, two-points Guass-Legendre integral method is used to simplify the balance equations, and then the iterative method of Newton and the steepest descent method are combined to solve these non-linear equations. Furthermore, the methods for measuring weight, drawbar pull, torque, slip and sinkage of wheel are discussed. In particular, a new method for visual wheel sinkage measurement for a lunar rover is proposed. The simulation experiment results show that the methods are effective and robust.In the last part of the dissertation, the estimation of a system of lunar soil mechanical parameters is constructed based on a lunar prototype rover and on-board sensors, and then an online experiment for the estimation of the terrain parameters is performed based on the proposed algorithm. The experiment results show the feasibility of the proposed terrain mechanical parameters estimation method.