Research On Indoor Mobile Robot Localization Algorithm Based On Multi-sensor Information Fusion

As the application field of indoor mobile robots continues to expand,the localization accuracy of mobile robots is becoming more and more demanding.Based on the mobile robot localization algorithm,this article decides to adopt a multi-sensor information fusion method for the problem that it is difficult to guarantee the localization accuracy by single sensor data localization,and also considering that the localization algorithm in a specific environment can affect the system localization accuracy.The localization algorithm is improved,and the accuracy and feasibility of the localization algorithm are verified through experimental simulation and result analysis.The main research contents of this article are as follows:(1)Establish the motion model and observation model of the mobile robot to predict and correct the position of the mobile robot.The hardware module with hierarchical layout of the mobile robot is built,and the robot operating system(ROS)is used as the basis for topic publishing and subscription of the sensor module to realize data transmission.Experimental and principle analysis of multiple localization algorithm fusion levels and methods are conducted,and specific methods of multi-sensor fusion are proposed by combining individual sensor characteristics and indoor application scenarios.(2)To address the problem of a single sensor acquiring a small amount of data and a large cumulative error during the motion of a mobile robot.By analyzing the sensor characteristics,the Extended Kalman Filter(EKF)based algorithm is proposed to fuse the sensor data of odometer,Inertial Measurement Unit(IMU)and Li DAR.The prediction equation is established using the position of the odometer and the attitude angle of the IMU to predict the robot’s pose,and the data obtained from the Li DAR are used to establish the observation equation to correct and update the predicted pose.The experimental simulation analysis of the collected data shows that the errors of the horizontal and vertical coordinates and the attitude angle are improved by 11.1%,22.8%and 22.9%,respectively,compared with those of a single sensor after the fusion based on the EKF algorithm,and the positioning accuracy is improved.(3)To further improve the localization accuracy of the mobile robot,the Adaptive Monte Carlo Localization(AMCL)algorithm using Kullback-Leibler Divergence(KLD)to sample particles is used.The predicted particle set of AMCL algorithm is obtained by varying the sampling of particles and sampling the fused poses of the EKF algorithm,and then updating the particle set using the point cloud information from the Li DAR scan.Through experimental simulation analysis of the data collected in the environment,the fused AMCL algorithm-based horizontal and vertical coordinates and attitude angle errors are improved by 20.7%,24.3%,and 23.2%,respectively,compared with the Gaussian Newton Optimization(GNO)algorithm,enabling the mobile robot to achieve positioning accuracy within 15 cm.To further verify the stability and robustness of the improved AMCL algorithm fusion,1200 poses in the fusion algorithm trajectory were analyzed,and the results showed that all the poses in the AMCL algorithm-based trajectory were of good accuracy.