Localization of Suborbital Vehicle Using On-Board Map and Camera Image

Abstract: Recently, unmanned aerial exploration of planets is an emerging field of interest for researchers all over the world. One of the major challenges that lie in the exploration aircraft is its localization in an environment where there is no useful navigation system like GPS. Although several localization algorithms were proposed, like SURF and SIFT, they suffer from premature convergence in an unknown and harsh terrain atmosphere (using satellite image mapping). The aim of this research is to develop a localization algorithm to detect the optimum features from the video, obtained onboard, and performs fast feature mapping of the features with the satellite image. The features are then mapped to the exact location on the database image and the path of the kite plane is obtained on the database image using RANSAC. The localization algorithm was evaluated in flight experiments using a kite plane. The experimental result of the algorithm showed localization of the aircraft using OpenCV (using CPP Console). The experimental results showed that the proposed algorithm (LAIGDE) outperforms the traditional mapping techniques like SURF with high accuracy and robustness.

Keywords: SURF, RANSAC, Histogram, GPS, Homography, Mapping.