Mobile Robot Based On Lidar And Binocular Vision Research On Autonomous Navigation Technology

Simultaneous Localization and Mapping(SLAM)is the hot spot and difficult area in the research field of mobile robotics and the prerequisite for mobile robots to achieve autonomous navigation in unknown environment.At present,the twodimensional state of the mobile robot SLAM problem solving method is mainly based on the extended Kalman filter method and based particle filter method,but the EKF-SLAM method exists when the first-order linearization of the nonlinear system exists Linearity error,and with the increase of the number of road signs,the problem of dimensionality arises.The RBPF-SLAM method still has the problem that the signpost observations are EKF and particle filter particle deficiency.These will eventually cause the map constructed by the mobile robot to be inaccurate and robust.Not strong and so on.Aiming at the current problems,this paper mainly studies the problem of synchronous positioning and map construction(SLAM)of tracked mobile robots.Synchronization positioning and map building(SLAM)of crawler-type mobile robots are the main research contents in this dissertation.Firstly,the research status and trend of mobile robot synchronization positioning and map construction are discussed and summarized;various models and experimental research platforms of mobile robot system are analyzed.Secondly,aiming at the problem of can not locate in time,inaccurate and robustness is not strong when build map with a single sensor when mobile robot's autonomous navigation in unknown environment,a integrate navigation method is proposed based on information fusion of binocular vision and laser lidar,landmark database containing the world coordinate system is establishedand the global map is obtained to realize autonomous navigation.Thirdly,for the general improvement RBPF-SLAM,the odometry motion model is taken as the proposed distribution and the odometer motion model is relatively noisy,resulting in a large difference in the weights among particles,an improved RBPF algorithm is proposed to combine the observed data(visual information and laser radar information)of the mobile robot with the mileage Information fusion,the number of required particles is effectively reduced and the uncertainty of the mobile robot in the prediction stage of particle filter is reduced.In view of the problem of slow extraction of feature points in general visual images,the visual images are processed based on the ORB algorithm to accelerate the speed of visual image processing.Lastly,the effectiveness and feasibility of the method is verified by experiments of crawler-type mobile robot with an Open Source Robotic Operating System(ROS).Experiments show that more reliable and accurate 2D raster image could be constructed by redundant information provided by the multi-sensor fusion and the robustness of mobile robot SLAM is improved.The research results of this paper provide a new idea for the study of mobile robot SLAM,which lays the foundation for the autonomous navigation research of mobile robots and has certain engineering practice significance.