Attitude Estimation Algorithm and Verification based on Binocular Vision of Micro Aerial Vehicles

摘要

Aiming at the navigation of Micro Air Vehicles (MAVs) in the natural environment, an algorithm is studied to measure the attitude of MAVs based on binocular vision. Binocular images are taken by calibrated binocular camera. The image feature points are extracted by Harris algorithm, feature points are matched with Normalized Cross Correlation (NCC) algorithm, and obtained a set of matching points. The three dimensional coordinates of the feature points can be calculated in camera coordinate system. Then the attitude of MAVs is estimated by Singular Value Decomposition (SVD) method and Least Median Squares (LMedS) theorem. The experiment result shows that the attitude estimation error is small, which can meet the requirements of MAVs navigation.