Filtering Algorithms For Positioning And Mapping Of Mobile Robots

The birth of mobile robots marks that humanity has entered the era of artificial intelligence.Mobile robots with autonomous navigation function represent the frontier of development in the field of robotics.The autonomous navigation of mobile robots is inseparable from synchronous positioning and map construction technology,namely SLAM technology(Simultaneous Localization and Mapping).The core idea of SLAM technology is positioning and mapping.Specifically: in an unknown environment,the mobile robot estimates its own position and direction,and uses sensors to collect information on the surrounding environment to complete the map construction,and achieves positioning and mapping.Nowadays,the application environment of mobile robots is becoming more and more complicated,and there are more and more disturbances and uncertainties in the external environment.Therefore,how to further improve the positioning and mapping accuracy of mobile robots,it has certain value to promote the development of mobile robots in the field of autonomous navigation.The research of this paper is mainly divided into the following aspects:Firstly,This paper describes the mathematical model of the SLAM problem,and analyzes the SLAM system model,establishes the coordinate system model,motion model and observation model of the mobile robot,discusses the representation method of the environment map by category,and selects the appropriate environment map representation according to the subject method.Secondly,This paper deeply studies the EKF-SLAM algorithm.In view of the low estimation accuracy of the Extended Kalman Filter,susceptibility to noise interference,and error accumulation,it innovatively proposes NNEKF-SLAM algorithm and MI-NNEKF-SLAM algorithm,which combines the principle of innovation and neural network,improves the EKF-SLAM algorithm for nonlinear systems by training the neural network to predict,update and compensate for system errors,and combines the multiple innovation principle The state estimation accuracy enables the Extended Kalman Filter to be expanded to multiple innovations when using a single innovation,thereby improving the system state estimation accuracy.The effectiveness of the improved EKF-SLAM algorithm is verified by simulation.Finally,This paper deeply studies the FastSLAM algorithm,analyzes the problems existing in the FastSLAM algorithm,and innovatively proposed an improved FastSLAM algorithm(TPSO-FastSLAM),through the interaction of particles in the particle swarm,the number of particles required by the algorithm is reduced.At the same time,the elimination constraint is introduced to discard the particles with a smaller adaptation value,which reduces the calculation amount of the algorithm and avoids particle exhaustion in particle filtering,alleviate the problem of particle depletion,and improve the estimation accuracy of the algorithm.The effectiveness of the improved FastSLAM algorithm is verified by simulation.