Research On Simultaneous Localization And Mapping Of Mobile Robot

In recent years,as a popular basic research direction,mobile robots have received widespread attention.Simultaneous Localization and Mapping(SLAM)technology is the key to autonomous navigation of mobile robots.When mobile robots fail to locate in an unknown environment Provide necessary support.Therefore,this paper starts from the two aspects of probability-based Kalman SLAM and data association in SLAM,and studies the synchronous positioning and mapping technology of mobile robots.The main contents are as follows:(1)Starting from the mobile robot SLAM,the mobile robot kinematics model,sensor observation model,environment map model and data association model are established,the coordinate system required in the implementation process is defined,and the mobile robot is established and explained based on the above model.Robot SLAM probability model.(2)In order to solve the poor consistency in the operation of the particle filter-based simultaneous localization and mapping(Fast simultaneous localization and mapping,Fast SLAM)algorithm,and the estimation accuracy of the particle filter is gradually degraded and depleted in the process.To reduce the problem,this paper proposes a Fast SLAM algorithm that improves the particle suggestion distribution function and resampling.In this algorithm,the strong tracking square root central difference Kalman filter(Strong Tracking Square Root Central Difference Kalman Filter,STSRCDKF)is used in the mobile robot positioning estimation stage to obtain the adaptive adjustment distribution function,making it more similar to the particle posterior probability distribution,and then Improve particle sampling accuracy;In the map estimation stage,STSRCDKF is used to determine the location of environmental road signs and improve the accuracy of mobile robot mapping.Finally,in the resampling stage,an adaptive bat heuristic resampling technology(BAR)is proposed.Which overcomes the problems of particle degradation and depletion,and has a better optimization effect than the existing SR,PRR and AGR methods.The system re-sampling strategy is performed on the particles,so as to ensure the synchronization of the particles while reducing the phenomenon of particle degradation.Diversity.The simulation results show that,compared with the STSRCDFast SLAM algorithm and the Fast SLAM2.0 algorithm,the algorithm proposed in this paper reduces the consistency deviation,improves the accuracy of real-time mapping,and has more robust performance.(3)Data association is used as a necessary prerequisite for state estimation in mobile robot SLAM to ensure its convergence in the process of positioning and mapping.Aiming at the problem that the data association algorithm used in SLAM cannot balance the computational complexity and the degree of association,this paper proposes a joint maximum likelihood data association algorithm.The SLAM data association problem is transformed into a combinatorial optimization problem,and the bat is used to search for the optimal solution for data association.The simulation results show that the data association algorithm proposed in this paper ensures the convergence during its operation,simplifies the complexity of the original algorithm,and improves the degree of association.(4)Build a mobile robot platform,and perform simulation comparisons of three algorithms in three environments: indoors,corridors and circular corridors.In the indoor test,the IBFast SLAM algorithm proposed in this paper has better boundary definition,detail completeness and particle suppression effect.In the corridor test,the IBFast SLAM algorithm proposed in this paper is better in detail completeness and particle suppression effect;in the 100-meter error,the root mean square error control deviation is smaller.In the closed-loop corridor environment,it can be seen that the BAJML algorithm proposed in this paper has a better looping effect,and the rest of the JML algorithm has a slight deviation from the JCBB algorithm at the loop.