Reserch On Navigation Algorithm For Soft Landing Asteroid Based On Images Sequence

Autonomous navigation is one of the key techniques for deep space exploration, which can reduce operation complexity and lower mission cost. In this paper, image based autonomous navigation algorithms are studied deeply for the asteroid landing phase.First, the elements of the image navigation are analyzed and the coordinate systems are defined. With projection relation of perspective projection and epipolar geometry considered, essential matrix is derived. The key techniques for navigation are discussed, which is the academic foundation of the sequential research.Second, feature point detection and tracking form navigation image sequence is studied. Image pre-process is performed by use of median filter. The model of feature tracking window is built up, then, KLT feature point detection and tracking algorithm is given out and feature point necessary conditions are discussed. The simulation results show that the feature point detection and tracking algorithm is feasible.Next, matched points based essential matrix estimation is studied. The spacecraft attitude and position parameters are derived from essential matrix and the scale of motion is recovered with range information from laser ranger taken in account. Random block feature point selection is developed to avoid feature point close clustering.Finally, image based asteroid lading navigation system structure is designed and numerical simulation is performed. The framework of the image navigation system is designed and functions of every subsystem are introduced. The simulation results suggest the navigation system proposed in this paper is feasible.