Research On Navigation Solution Of Multi-rotor UAV Based On Multiple Sensors

Due to the characteristics of portability,flexibility and strong function,small multirotor unmanned aerial vehicles(UAV)have aroused great concern of the public in recent years.The attitude control system of UAV needs to receive accurate attitude information as feedback to control UAV.Meanwhile,if the UAV wants to achieve the functions of trajectory tracking,fixed hover,etc.it also needs the accurate location and velocity information.However,there is limited space in the UAV,so the navigation sensors that it carries are more of small sensors,which has the characteristics of low precision and large measurement noise.Therefore,just depending on the measurement results of a single sensor can't gain the accurate navigation data.Based on the design of data fusion algorithm,the research emphases of the thesis is to fuse the measurements of multiple sensors and restrain the noise of single sensor,thus gaining more accurate information of attitude,location and velocity,Therefore,the study of the thesis on navigation solution of small multi-rotor UAV is of great research value and application prospect.Firstly,the thesis briefly introduces the knowledge of navigation related principles and the characteristics of frequently-used navigation sensors on the UAV.The contents mainly include common navigation coordinates,the representations of attitude angles,measuring principle and noise characteristic of gyroscope and accelerometer as well as attitude measuring principle of magnetometer and accelerometer.Secondly,aimed at the problem that the high frequency vibration caused by turning motor in the flying process makes the measured value of gyroscope and accelerometer have large deviation,the thesis researches and designs digital low-pass filter to filter out the high frequency noise caused by body vibration among sensor output signals.Therefore,this thesis analyzes the influence of the body high frequency vibration on the output signals of two sensors to get frequency characteristics of noise.Then it uses hamming window function method and yulewalk function method to respectively design FIR digital low pass filter and IIR digital low pass filter.The author takes filtering effect and filtering real-time performance into consideration and designs the digital low pass filter which can effectively restrain the amplitude of body high frequency noise.Thirdly,in allusion to the problem how to accurately acquire the attitude information of UAV,the thesis designs two-order complementary filtering attitude algorithm in quaternion form and verifies the calculating effect of the algorithm.According to the principle that the frequency characteristics of gyroscope,accelerometer and magnetometer are complementary,the thesis uses the gyroscope as the main measuring element and the accelerometer and magnetometer to correct the results,thus gaining good attitude calculation results.Meanwhile,the paper compiles algorithm routine and sets up the simulation system in Matlab,which respectively uses simulation data and experimental flight data to verify the algorithm in Matlab.Finally,aimed at the problem how to accurately gain the location and velocity of UAV,the thesis designs the arithmetic which is based on the expansion of location and velocity of Kilman filtering and verifies the arithmetic at the same time.The arithmetic makes use of accelerometer,barometer and GPS to measure the location and velocity in the flight process of UAV and uses Kilman filtering algorithm to fuse the information of the three measurements.Based on continuous-discrete time system,the algorithm introduces the offset of the accelerometer as state vector,which makes the information through the algorithm more accurate;meanwhile,the thesis uses the fourth-order Runge-Kutta method to avoid calculating Jacobian Matrix and improve the calculating speed.Then the thesis takes advantages of simulation and experimental data to verify the algorithm as well.