Localization Of Mobile Robot Based On Multi-Sensor Data Fusion

Multi-sensor fusion based on localization method is the mainstream approach of mobile robot localization at present.Compared with single sensor localization,multi-sensor fusion localization has the advantages of low cost,high error tolerance,and strong anti-interference capability.However,the existing multi-sensor fusion localization algorithms cannot simultaneously take into account localization accuracy and system fault tolerance,and less consideration is given to information distribution,signal interference,and error model anomalies in the fusion process,which seriously affects the robot localization effect.To address these problems,this thesis focuses on the principle of sensors,the federated filter information distribution coefficients and adaptive robust federated filter,and the main research contents include.Analyzing the basic principles of sensor navigation and localization for mobile robots,deriving the conversion relations of commonly used navigation coordinate systems,establishing the error models of inertial guidance,GPS and odometer,analyzing the advantages and disadvantages of each,and establishing a combined navigation and localization framework based on the complementary characteristics of sensor advantages and disadvantages.A new information distribution method is proposed for the problem of localization accuracy degradation due to subsystem failure in the traditional federated filter structure.Considering the system state covariance matrix and observable matrix characteristics,the overall distribution coefficient of the subsystem is calculated based on the system state covariance matrix;the state variable observable matrix is used for the secondary distribution of the distribution coefficient,so that the information distribution coefficient can be changed according to the system state change and the localization accuracy of the main system after fusion is improved.The simulation results show that the optimized information coefficient can effectively improve the localization accuracy of the combined navigation system.A new adaptive robust localization algorithm is proposed to address the problem of sensor degradation of localization accuracy due to complex environment and signal perturbation.Based on the optimized information distribution coefficient,the state noise model is estimated in real time,and the adjustment coefficient is introduced to continuously update the measurement covariance matrix to improve the online estimation of uncertain system noise and coarse estimation ability,and improve the localization accuracy of the system.Simulation results show that the algorithm can remove sensor coarse and estimation uncertainty noise and improve the localization accuracy of the navigation process.