| It is acknowledged the first time for China to leave footprints on the moon that the moon’s lunar rover successfully land on the moon and parade on the lunar surface to carry out exploration work,which marks our country start the exploration of unknown environment of planet’s surface. SLAM(Simultaneous Localization and Mapping) combines the localization and mapping of mobile robot,which is recognized by academics as the key point and foundation of achieving authentic autonomy of mobile robot universally.Incremental environment map can be built and robot self-localization achieved according to pose estimation and environment perception in the motion, which can be better used in the field of localization and mapping of robot in unknown environment. Visual sensor can obtain intuitive and rich information of environment. It also has the merit of low cost, mature technology,strong environment adaptability,which is used widely in the field of mobile robot.Therefor it is very important theoretically and practically to research on the methods of simultaneous localization and mapping for vision-based mobile robot.For the localization and mapping of mobile robot in unknown environment, SLAM system model and EKF-v SLAM algorithm framework monocular vision-based is established. Meanwhile,a mobile robot monocular vision-based is developed for the follow-up test which provide a proof for related theories and algorithms through simulation and experiments. Firstly,This thesis summarizes the history and research status of SLAM and vision-based mobile robot.Secondly,basic theory,technical difficulties,algorithms and common mapping methods of SLAM is studied.Thirdly,the model of SLAM based on monocular-vision mobile robot is built,which includes motion model,cameral model,observation model and EKF-v SLAM algorithms.Fourthly,test platform for v SLAM is built which include mechanical structure,control system,motion system,camera system,communication system and so on.Meanwhile, error of odometer and signposts observation is studied through practical experiments. Finally,in order to validate the model and algorithms proposed,experiment and simulation analysis based on issue arrangement is conducted, which includes the simulation analysis of KF and EKF,calibration experiments of camera parameter, EKF-v SLAM experiments in 2D and 3D environment, simulation analysis of the relation between the quantity of landmarks and system performance,simulation analysis of the relation between the motion path of robot and error canceling.It is verified by the experiment and simulation analysis that the good effect of KF in linear signal filtering and EKF in nonlinear signal filtering. Simultaneously, the EKF-v SLAM algorithms proposed in this thesis can constraints the cumulative error and reduce the uncertainty of localization and mapping for mobile robot in unknown environment which improve the autonomous of mobile robot to some extend. |
PDF Full Text Request