The Attitude Estimation Fusion Methods And Navigation Methods Of Micro-rotor Drone

Attitude and Navigation System is the crucial component of UAV(Unmand Aerial Vehicle) system.For Small UAV, with the restrains of payload and manufacturing cost price, the attitude and navigation device should be light weighted and low costed. Thanks to the rapidly developing of MEMS(Microelectromechanical System) technology, the low cost inertial chip(like accelerometer,gyroscope) can be widely used in UAV’s attitude and navigation system. Generally speaking we call the module that can measure drone s attitue as IMU(Inertial Measurement Unit),which include three-axis gyroscope and threeaxis accelerometer. AHRS is short for Attitude and Head-ing Reference System, which usually has three-axis gyroscope, three-axis accelerometer and three-axis magnetometer,can provide heading information. So we can call IMU as six-axis attitude module, and can call AHRS as nine-axis attitude module. Based on the data from attitude module,altitude data from barometer, position and velocity data from GPS, one basic navigation system can be constructed. But the unit of sensors’ measurements is different from each other, a filter algorithm is needed to fuse those data. So in the thesis the fusion algorithm of attitude module and navigation module will be discussed.In attitude module, the process of fusion different MEMS sensors’ data is call attitude estimation. So the module’s performance depends on two parts: the performance of sensor chip and the performance of estimation algorithm. Till so far, various of estimation algorithms have been promoted and two main sets methods are used in micro rotor drone. The first set is based on kalman filter and its derivants, like MEKF(Multiplicative Extented Kalman Filter), UKF(Unscented Kalman Filter),etc. The second set is based on complementary filter, for example, linear complementary filter, gradient complementary and nonlinear complementary filter. Those filter methods both have advantages and disadvantages and can be choosed according to practical environment, like the rank level of MEMS chip and capability of onboard MCU. A lot flying data showed that estimation algorithm is only one part of variables which influence the reliability of AHRS, mechanical vibration and the external acceleration during flying can be the bigger one. A lot algorithms had been promoted with testing only under ideal environment, they had not analyzed the performance in flying condition; With external acceleration problem,some people came out with the idea which use GPS velocity information to complement and correct, but the precision is not good enough and update rate is low. So in this thesis it will(1)try to analysis the flying mode and reduce the impact of extern acceleration;(2)analysis those popular attitude estimation’s performance in different flying state;(3)base on specific quadrotor’s dynamic model try to find a new way to estimate extern acceleration.In navigation system, the first priority is to get position information. There exist a lot noise in position data and velocity data given by GPS, so in short period by a first and second order integration we can get more accurate velocity and position from attitude module’s acceleration output, after fuse those two kind of velocity and position more stable and accurate estimation navigation data can be obtained. Under some extreme circumstances, there is no GPS signal, with the consideration of improving navigation’s reliability and system redundancy more way needed to calculate position. Integrated above two aspects in this thesis it will(1)construct one algorithm based on kalman to fuse the data from GPS,barometer,attitude module to calculate velocity and position data.(2)construct a vision navigation platform, use airborne computer and camera to design a new navigation method based on SIFT and relative computer vision knowledge.