Abstract: The attitude of a lunar lander needs to be adjusted quickly during its descending. However, the traditional celestial navigation algorithm takes a long time to identify the star and occupy some hardware resources, which wastes power consumption and limits the dynamic response ability of the lunar lander. To solve this problem，a new method based on image gray error is proposed. This method uses the INS data to build the image gray error function and optimizes the attitude along the gradient of image gray error, which can estimate the attitude without star identification. The simulation results show that, when the number of navigation stars is no less than 3, the INS/CNS deeply integrated navigation algorithm can reduce the time by 60%, and the attitude accuracy is maintained within 10″ (non optical axis) and 50″ (optical axis). This method can still work when the number of navigation stars is less than 3, and the attitude accuracy is maintained within 50″ (non optical axis) and 100″ (optical axis). The INS/CNS deeply integrated navigation algorithm helps to reduce the calculation, and save hardware resources and power consumption, which can provide theoretical reference and new ideas for the future design of a lunar lander navigation system.

Key words: Celestial navigation, Inertial navigation;Nonlinear optimization;Integrated navigation