Research On Mobile Robot Location Based On Improved Particle Filter

In recent years,with the development of artificial intelligence,mobile robots are widely used in various fields.Location problem is the basic problem of mobile robot research,the premise of autonomous navigation,and the foundation of robot successfully completing tasks.Particle filter algorithm has been widely concerned by scholars because of its adaptability in non-linear and non Gaussian systems.Monte Carlo positioning method based on particle filter algorithm has been successfully applied to mobile robot positioning,but there are some defects.The standard particle filter has the problems of particle degradation,lack of diversity and adaptive particle number.In this paper,the basic particle filter algorithm is improved and applied to the positioning of mobile robot.First,using the root mean square embedded volume Kalman filter to generate the proposed distribution,the current observation information is integrated into the algorithm,so that the proposed distribution is closer to the actual posterior probability distribution of the system,and to some extent,the case degradation phenomenon is solved.Secondly,for the lack of particle diversity caused by resampling process,chopthin sampling instead of basic resampling process is no longer a process of high weight particles replication and low weight particles elimination,but a process of producing particles with different weights.At the same time,chopthin sampling can be carried out in every iteration period.Compared with the basic resampling,the effective particle number is more stable.Third,the standard particle filter algorithm needs a large number of particles to get more accurate estimation results,which has a large amount of computation and affects the real-time performance of the algorithm.In order to adjust the number of particles adaptively,this paper calculates the JSD to determine the similarity between the subset of particle set,and adjusts the total number of sampling particles online.Finally,the simulation results show that the improved algorithm has better localization performance.