Simultaneous Localization And Mapping Algorithm For Mars Rover Based On Filtering Theory

With the rise of a new round of Mars exploration activity,the major space powers are competing to carry out the surface exploration activities of Mars,and the Mars rover is the key to complete this task.Simultaneous Localization and Mapping is also known as SLAM,which refers to positioning a mobile target realitively in an unknown environment and simultaneously building a map of the environment.In the absence of global positioning information on the surface of Mars,SLAM algorithm can localize the rover relatively as well as building a map of the environment,so as to support other tasks such as path planning and obstacle avoidance.In this context,this thesis studies the SLAM algorithm based on the filtering theory,and he main research contents are as follows:Firstly,the mathematical model of SLAM algorithm is studied,and SLAM is modeled by probabilistic method.Considering that the rover moves in a threedimensional scene,it is necessary to establish the three-dimensional model of the rover system,and study the contact constraint between the rover and the terrain surface.Then a simulation is needed to obtain the motion parameters of rover satisfying the terrain constraints,so as to lay the foundation for the subsequent research of SLAM algorithm.The extended Kalman filter is used as the core estimation algorithm to estimate the location and attitude of the rove as well as the location of landmarks in the environment.The main flow of the EKF-SLAM algorithm is studied,and the algorithm is simulated to verify its feasibility.Considering that the computational complexity of EKF-SLAM squarely grows with the increase of the number of landmarks,the Rao-Balckwellised particle filter based SLAM algorithm of Mars rover is studied.This algorithm decomposes full SLAM into a series of low-dimensional estimation problems,which can be solved using Rao-Balckwellised particle filter,so that the computational complexity is reduced.Finally,the feasibility and validity of the algorithm are verified by mathematical simulation.In traditional SLAM,the robot carries out SLAM passively.Considering this,this thesisi studies the active SLAM algorithm of the rover by planning the robot movement to reduce the uncertainty of localization and mapping.On the basis of EKF-SLAM algorithm,the mathematical model of active SLAM is established,and a solution of active SLAM based on model predictive control theory is studied.The validity of this method is compared with the traditional SLAM algorithm in reducing the uncertainty of localization and mapping by mathematical simulation.