The Research And Application Of Nonlinear Filtering In Mobile Robot Attitude Measuring System

With the development of science and technology,the accurate acquisition of mobile robot attitude information has become the first condition to realize the function of robot autonomous navigation and precise control.Mobile robot attitude measurement has become a hotspot in the field of robotics.There are many non-linear links in the mobile robot attitude measurement,this paper uses the quaternion method to establish mobile robot nonlinear attitude filtering model.Aiming at the shortcomings of common non-linear filtering algorithms under this model,this paper designs an attitude filter based on UPF.An improved attitude filtering algorithm is proposed to eliminate the interference of motion acceleration and random magnetic on the measurement accuracy.A low power combination scheme of INS/Magnetometer is used to design the embedded system for experimental verification.Firstly,this thesis investigates the research status of nonlinear filtering algorithms at home and abroad and the application status of mobile robot attitude measurement technology.In this paper,the advantages and disadvantages of six common attitude measurement schemes are analyzed,and the attitude measurement scheme of the inertial navigation / magnetometer is determined.The attitude measurement module of the system is composed of MEMS triaxial accelerometer,MEMS triaxial gyroscope and MEMS Three-axis magnetometer composition.Secondly,the attitude sensor error model and the relationship between the attitude sensor and the robot attitude angle are established.The filtering performance of EKF in quaternion filter model is analyzed by simulation.The filter based on UPF is designed for the defects which are affected by initial error and Gaussian conditional constraints.Aiming at the influence of motion acceleration and random magnetic interference in the operation of mobile robot,an improved attitude filtering algorithim based on UPF is designed.Then,by setting different simulation environment,the performance of UPF based quaternion model under non-Gaussian condition is studied and compared with UKF and PF,and the contrast parameter index is given.By comparing the UPF,EKF and UKF under different initial error conditions,it is verified that the designed UPF filter has little effect on the large initial error.The improved UPF attitude filter is verified by simulation.Finally,an embedded attitude measurement system is designed and the improved UPF filter algorithm is experimentally verified.According to the design program to determine the main sensor model,this paper designs the functional module,PCB and completes the experiment.Simulation and experimental results show that the improved attitude filter based on UPF can reduce the interference of motion acceleration and magnetic on the attitude measurement of mobile robot to a certain extent.