| Nowadays,with the rapid rise of the artificial intelligence industry,how to accurately obtain posture information has become the key to the control of unmanned systems.With the continuous development of inertial technology and the iterative update of information fusion algorithms,the multi-sensor fusion attitude solution makes up for the large error of the traditional single sensor.High precision,low cost,and easy portability have become synonymous with attitude measurement systems,but most of the existing products contradict the supply and demand relationship in the low-consumption market.Aiming at the above phenomenon,this topic will study some key issues in the attitude calculation of unmanned systems.First of all,in view of the contradiction between cost and accuracy in the attitude measurement system,the subject takes the high-performance microprocessor STM32F407ZGT6 as the core processing unit of the system,and the domestic MPU9250 chip integrating gyroscope,accelerometer and magnetometer as the main attitude perception unit to design an attitude measurement system.Based on the arm embedded system,the task scheduling and management of attitude sensing data acquisition and processing,error compensation,attitude update,solution fusion,and data communication are realized.Secondly,in view of the problem that the low-rate IIC communication mode of the attitude sensor in the past may cause the delay of the attitude calculation in the later stage,the subject switched to the SPI high-speed serial communication bus to increase the sampling frequency.Aiming at the complex problem of traditional sensor calibration methods,a simple and efficient turntable calibration method was designed.The error model is established by analyzing the characteristics of each sensor,and the error parameters are identified and calibrated by analytical method,Kalman filter and ellipsoid fitting respectively.Then,the current popular attitude calculation methods are researched and designed,and the factors affecting the calculation accuracy are analyzed through experimental comparison.Based on this,a random interference detection mechanism is established to adapt the observation noise covariance matrix of the accelerometer and magnetometer;at the same time,the random error of the gyroscope is filtered,and the ASUKF attitude calculation method is designed to improve the calculation accuracy.The random error of the gyroscope is analyzed,the measurement residual training model is established,and the Gaussian process regression is used to train the random error of the gyroscope to reasonably estimate its mean and covariance,and further improve the solution accuracy.Aiming at the problem that the engineering method of calling the matrix operation library affects the real-time performance of the solution,the subject adopts the model reverse order output method to engineer the multi-sensor fusion attitude calculation algorithm to improve the real-time performance of the system solution.Finally,a low-cost multi-degree-of-freedom turntable is built,and SPWM is used to drive the turntable motor to simulate the pose transformation of the unmanned system,and the attitude information is collected by the angular displacement sensor.Based on the designed attitude measurement system and test turntable,experiments are designed to verify the feasibility of the designed sensor calibration method.At the same time,by comparing with the high-performance attitude reference system HWT906,the attitude calculation method designed by the subject is analyzed and compared from the static and dynamic joint experiments,and a dual-mode attitude calculation and measurement system is established to reduce unnecessary power consumption and ensure measurement precision.The joint static and dynamic experiments have confirmed that the attitude measurement system designed by the subject and its matching attitude calculation method have good robustness,which is in line with the research expectations. |
PDF Full Text Request