Research And Implementation Of Location Algorithms For Unmanned Vehicle

Simultaneous localization and mapping(SLAM)is a research main issue in the autonomous navigation field of unmanned vehicles.FastSLAM strategy further decomposes the problem into "localization" by using particle filtering and "mapping" by using extended Kalman filtering,where the reliable localization is the key for autonomous mobile systems.In this paper,the fast localization problem of unmanned vehicles is focused on,and the FastSLAM method based on particle filter is specially studied.At the same time,in order to get the models necessary for the FastSLAM algorithm and to eliminate the inaccurate localization problem in the traditional FastSLAM method,several improvement algorithms are proposed and the relevant simulation experiments are carried out based on MATLAB.The main research contents include:First,a Particle Swarm Distribution Estimation(PSDE)strategy based on the real-time evolution state is proposed,and the criterion for determining the real-time evolution state of the population is defined,while the PSDE-FastSLAM algorithm was established.At the same time,the particle diversity measurement index is introduced,and the optimal value of the parameter is obtained through simulation and applied to the subsequent analysis.Then,the double mutation operator(DM operator)is introduced to redefine the dynamic population distribution strategy and to enhance the search range near the mutant particles for overcoming the shortcomings of the PSDE-FastSLAM algorithm,while the PSDE-FastSLAM-DM algorithm is established.Simultaneously,the simulation experiment is designed for the double mutation rate and dynamic inertia weight,and the optimal value of the parameters are obtained and applied to the subsequent calculation.In addition,the effects of DM operator and other evolution operators on the algorithm are compared and analyzed,the simulation results proved the advantages of DM operator.Finally,two evaluation indicators,which are the path deviation and the operation time,are selected to design an unmanned vehicle simulation experiment based on the PSDE-FastSLAM-DM localization algorithm,moreover it is compared with various algorithms.The results show that the proposed PSDE-FastSLAM-DM algorithm has good positioning accuracy and computing efficiency.