Research On SLAM Of Moblie Robots

The application level and studies degree of mobile robotics represents the level ofdevelopment for a country’s industrial automation, and it means an important strategicsignificance for national defense, social and scientific technology. Simultaneous Localizationand Mapping(SLAM) is the key factor for mobile robotics to achieve fully autonomousnavigation, and is a hot and difficult problem in the field of robotics research. So it hasattracted great attention from many researchers.According to the basic framework of SLAM and the related model of mobile robotics,this paper research the key issues about self-localization of robotics and real-time building ofthe environmental map. In order to enhance estimation accuracy, consistency and efficiency ofrobot localization and map building, the paper proposes some improved SLAM algorithms.Main contents of the paper are summarized as following:Firstly, we in-depth research the basic model of the mobile robot, and combine with thebasic framework of SLAM to build simulation platform. In this paper, the all simulationexperiments are finished on this platform. It lays the theoretical foundation for future researchwork to build the platform.Then, we introduce the conventional Kalman filter theory, and deduce theimplementation procedure of EKF-SLAM. To verify the feasibility and effectiveness of theEKF-SLAM, it is applied to the above simulation platform, and simulation results show thatthe proposed method is feasibility. When the EKF are applied to mobile robotics which is anonlinear systems, it can obtain the truncation error to calculate of Jacobi matrix. In order toovercome this defect, the improved method of SLAM which is based on UKF is presented,and is used in the SLAM simulation platform. Simulation results show that UKF-SLAM hasbetter estimated effect and stability than the EKF-SLAM, when the amplitude of noise areincreasing.Finally, In order to solve the particle degradation and particle dilution problems ofconventional FastSLAM, and improve the particle distribution, this paper presents a particlefilter SLAM algorithm based on adaptive artificial physical optimization. Through the virtualforce model of artificial physical optimization, we suppose that between the particle have theattraction and repulsion. the attraction impel the particles to approach posterior probabilitydensity, and solve particle degradation. Under the influence of repulsion, particles containeach other to avoid the duplication and excessive concentration of particles, and increase thediversity of particles. This algorithm improve the adaptive resampling algorithm which isaccording to degradation speed of particle after particle distribution. It is verified theeffectiveness of this algorithm through the simulation experiments.