Research On Simultaneous Localization And Mapping Of Mobile Robots Based On Probabilistic Model

This paper focuses on the filtering problem of mobile robot SLAM(simultaneous localization and mapping)algorithm.We propose AEKF-SLAM algorithm based on the improved adaptive extended Klaman Filter and Wo-FastSLAM algorithm based on the weight optimization theory.In order to overcome the problem of filtering divergence caused by model mismatch,our paper proposes a method to adjust the gain size of Klaman Filter by using the enhanced memory and fading memory,which improves adaptive extended Klaman Filter,and an AEKF-SLAM algorithm based on adaptive extended Klaman Filter is designed.This method changes the weights of the old observations and new observations in the extended Klaman Filter in order to avoid the EKF-SLAM filtering divergence and improve the accuracy of mobile robots in autonomous positioning and mapping.Aiming at the problem of particle degradation and diversity attenuation in the traditional FastSLAM algorithm,we put forward a method of using weight optimization theory to select the particles which have lager weight based on Rao-Blackwellised particle filter.A WO-FastSLAM algorithm based on the theory of particle weight optimization and the frame work of SLAM theory is proposed.It optimizes the map building while updating the state information of the robot.Moreover,the proposed method can ensure the diversity of particles,and effectively restrain the phenomenon of particle degradation,which improves the performance of SLAM algorithm.Finally,simulation experiment of the two algorithms is implemented on Matlab platform,and the results indicate that: the improved AEKF-SLAM algorithm of extended Klaman Filter effectively suppresses the filtering divergence problem and improves the accuracy of robot localization and mapping.Meanwhile,the WO-FastSLAM algorithm based on weight optimization theory effectively suppresses the attenuation of particle diversity and improves the performance of SLAM algorithm.