Application And Implementation Of EKF Filter For UAV Flight Control System Based On Real-time Operating System

The flight control system is the core of the UAV which controls the autonomous flight of the drone.The UAV is required to be able to fly stably in accordance with the prescribed route,first,you need to obtain the attitude information,and then you can determine the control amount to adjust the flight attitude.The acquisition of attitude as the primary goal and mission of flight control is crucial in the related research of flight control.The acquisition of the attitude relies on the attitude measurement system,which includes the sensor and attitude solving algorithm.Due to its small size,low cost and power consumption,UAV usually choose micro-electro-mechanical(MEMS)inertial sensors,such as three-axis gyroscopes,three-axis accelerometer,etc.,a single inertial sensor cannot get stably and reliably information for a long time,so it is necessary to add other auxiliary sensors,such as magnetometers,to form a multi-sensor attitude measurement system.For multi-type data provided by multiple sensors,the filtering algorithm is used to fusing data and estimating the optimal attitude,and provide a good input for subsequent attitude control.After analyzing and comparing the commonly used filtering fusion algorithms,this paper chooses Kalman filter(KF)and extended Kalman filter(EKF)which are small in data storage,excellent in estimation effect and widely used in engineering,to integrating data from each sensors and estimating states.In order to improve accuracy and long-term stability of the UAV,the gyroscope,accelerometer and magnetometer are combined to form a multi-sensor measurement unit,and the original data error from each sensor is modeled and calibrated.This paper designed a 13-order EKF attitude estimator to integrate the 9 data from above three sensors and estimate the 3 attitude angles displayed on the ground station.Contrasting experimental data,proved that EKF can fusing information effectively and steadily,the attitude angle error are kept within 2.135° on average,comparing with the calculating results based on a single gyroscope or accelerometer/magnetometer,results prove that precision can increased 73% and 33% respectively.In addition,in the view of the EKF's limitation which does not apply to the strongly nonlinear systems,EKF are preliminary improvement based on the multiple innovation theory,the simulation data shows that the mean square error of attitude angle decreased about 0.1 °,precision is improved.Finally,an extended study is carried out on the application of EKF in flight control,and a 9-order EKF was designed to integrate the accelerometer,GPS and air pressure altimeter 's data,and estimate the UAV speed and position information.The experimental results show that the 3d position error is kept within the 0.5m,and the 3d speed error is kept within 0.4m/s,which can provide good input for the speed and position control ring of the UAV,and ensure the UAV's fixed-point hovering and track tracking functions.In this paper,EKF is used to estimate the state information of UAV,including posture,position and speed.Moreover,the accuracy of the state information is better than calculation information based on single sensor,and the accuracy of the whole navigation information is improved by EKF.In view of the need for real-time and accuracy of attitude measurement,In this paper,the STM32F407 based on ARM Cortex-M4 F is selected as the micro-controller on the flight control hardware,the embedded real-time operating system FreeRTOS is selected as the software environment to utilize CPU resources more efficiently and reasonably,this can improve the real-time response capability of the system,therefore,the real-time performance of the calculation can be improved on the basis of the flight control software and hardware platform.