Position And Attitude Estimation Techniques Of MAV Based On Vision/mems

MAV has very broad application prospects in the field of military and civilian, since it has advantages of small size, light weight and low cost. Because of the characteristics of small size and light load, it has important practical significance that how to navigate for MAV by existing navigation system. Inertial navigation and vision-based navigation are integrated in the paper, base on MEMS inertial components and CCD camera, to estimate the position and attitude of MAV accurately during flying, by using the redundant measurement information provided full and effectively.The kinematic model is established in the paper. A flight path is designed and the simulations of the motion of MAV are made, based on SINS formed by MEMS inertial components. Simulations demonstrate that single and pure SINS will have accumulated error, leading to divergent results.Since the attitude can not be observed directly by SINS, the idea that integrate SINS and vision-based navigation is proposed in the paper. The keypoints of the video or image are extracted and matched first by using SIFT, based on the principle of motion estimation of MAV. The results of extracting and matching are tested. The position and attitude of MAV in geographic coordinate are estimated, by the relationship between the keypoints of images and ground environment, which can be used to locate initially. The relative position and attitude of MAV between the adjacent shooting time are estimated, by the keyponts of the two images photoed in the adjacent time, which can estamate the real-time position and attitude. Simulations show that estimation error is small enough to meet the requirements of vision-based navigation.The program of integrating the SINS and vision-based navigation is designed, considering their Advantages and disadvantages. The two types of data are fused by multi-rate Kalman filter, and the simulations are made, based on the flight path designed before. Simulations demonstrate that the results of multi-rate Kalman filter are better convergenced, compared to the single-rate Kalman filter, which can enhance the reliability of the integrated system.