Localisation method for a Mars aircraft using multispectral images

Abstract

In Japan, the working group for the Mars Exploration Aircraft continues to research and develop a Mars aircraft aiming to a future survey mission. To ensure the success of the flight mission, a self-localisation system with low-computational complexity is necessary. In the present research, a new self-localisation method is proposed using multispectral images. The algorithm is based on the simple mapping of image moment invariants and gradients calculated from several images at different wavelengths. The numerical simulations revealed the sufficient robustness of the proposed method to image noise. The estimation accuracy can be improved by increasing the number of the spectral images.