Attitude Estimation For Quadrotor UAV Based On MEMS Technology

In recent years,with the development of computer and electronic technology,four rotor unmanned aerial vehicle(UAV)develops rapidly.Because of its small size,simple structure and low cost,it can be used in military and civil fields,and has become the focus of people's research.When UAV working,we need to know the accurate attitude of UAV.Therefore,it is necessary to estimate the attitude information of UAV in real time.The traditional sensors does not meet the micro market demand because of the high cost and large volume.However,the MEMS sensor due to its advantages of small volume,low cost and small power consumption is widely used in the four rotor uav.In practice,there will be a drift caused by the MEMS gyro sensor which will cause attitude divergence and even cause the attitude divergence;MEMS accelerometer dynamic performance is not good,especially when the UAV accelerating.the body of quad-rotor will violently vibrate,which will affect the accelerometer in the form of white noise;The magnetometer is easily affected by the external magnetic field in the working process.These shortcomings of the MEMS sensor will lead to inaccurate acquisition of the attitude information of four rotor UAV.In order to solve the above problems,the attitude estimation algorithm should be designed to fuse the date measured by MEMS sensors.This thesis puts forward two kinds of attitude solution.The first one is the Extended Kalman Filter algorithm.In this algorithm,quaternion and gyro drift is used as state vector to establish the state equation.The acceleration and the magnetic field strength in the body coordinate system is used as observation vector to establish the observation equation.Finally,we use the extended Kalman filter algorithm to fusion the date.However,the extended Kalman filter algorithm's computing is relatively large and the requirements of the processor is relatively high,so the application in practice is relatively small.In view of the shortcomings of the extended Kalman filter algorithm,we propose the second kinds of attitude algorithms: gradient descent method.The algorithm firstly establishes quaternion differential equation and we use the measured data of gyroscope get the quaternion.Then,we can get the second quaternion by gradient descent method.Finally,we fuse the two groups of quaternion to get the best quaternion.Then the quadrotor UVA attitude information is obtained according to quaternion.The requirement of the algorithm is not high,and the computation is small,but the convergence rate is slow when approaching the optimal value.The following is the main content of this thesis:First of all,the quaternion of the background and significance is introduced.We introduce the current research status of four rotor UAV at home and abroad;introduces the research status of MEMS sensor at home and abroad;introduces the research status of the AHRS.Secondly,the basic knowledge of quaternion is introduced.We introduce the basic structure of the quaternion and makes a brief introduction to the principle of flight;We introduce quaternion coordinate system commonly used and man-machine coordinate system is established on quaternion UAV;We introduce two kinds of attitude repersentation methods: Euler angle method and four element method.Then,an extended Kalman filter attitude algorithm is presented to solve the attitude of quaternion.In this algorithm,quaternion and gyro drift is used as state vector to establish the state equation.The acceleration and the magnetic field strength in the body coordinate system is used as observation vector to establish the observation equation.Finally,we use the extended Kalman filter algorithm to fusion the date.The simulation results show that the proposed algorithm can solve the attitude problem and suppress the white noise.Finally,according to the shortcomings of the extended Kalman filter algorithm,a gradient descent algorithm is proposed to solve the attitude of quadrotor.The algorithm firstly establishes quaternion differential equation and we use the measured data of gyroscope get the quaternion.Then,we can get the second quaternion by gradient descent method.Finally,we fuse the two groups of quaternion to get the best quaternion.The simulation results show that the proposed algorithm can accurately obtain the attitude information of quadrotor.