Research On Mobile Robot Vision And Lidar Navigation System

Mobile robots are increasingly applied in the fields of intelligent manufacturing,warehousing and freight transportation,and automatic parking.Traditional mobile robots rely on external sensors for positioning and navigation,which are easily affected by the surroundings.And most of them can only move according to fixed trajectories.Mobile robots need to rely on their own sensors to identify their location in the environment and build an environment-based map.Then navigation is carried out based on the map.This technique is implemented by simultaneous positioning and building map(SLAM)method.This article is based on SLAM technology to design a mobile robot platform that can achieve navigation.This paper pays attention to robot chassis,hardware,robot chassis control,middle-level positioning navigation algorithm,and upper-level host computer communication.In bottom layer,the motor,encoder,driver,and robot wheel are chosen.The power system designed.And onboard computer of robot is selected.Furthermore,motion control system of the robot is designed based on the robot operating system(ROS).Motion model of the robot is calculated considering the virtual scene and actual situation.At the same time,the current navigation scheme is compared with the choice of lidar and vision sensor which are the main sensors for robot autonomous navigation.In the aspect of map construction algorithm,the laser radar and vision are used to construct the map.For the laser radar,the mathematical model of the laser radar is deduced firstly,and the laser radar SLAM process is deduced.In addition,the particle filter algorithm in the laser radar SLAM is deduced and compared.By comparing advantages and disadvantages of the popular laser radar SLAM scheme,one is chosen.The coordinate management tool in ROS is used to coordinate the change of the odometer information and carry out the algorithm migration.Considering that the visual SLAM scheme still has strong demand,this paper also adopts the visual SLAM scheme to complement laser radar SLAM.And the visual SLAM scheme mainly improves the ORBSLAM2.Two schemes were used to improve the ORBSLAM scheme.The scheme 1 improved it into a point cloud map.Due to the computational complexity and practicality,the scheme 2 was adopted,which was improved to a 2D raster map,and the relevant software packages in ROS were used.A 2D raster map was generated and a map was constructed.The navigation aspect is mainly based on AMCL and ROS navigation package.And the specific implementation of navigation principle derivation and navigation is carried out.The upper layer mainly uses remote communication to test and supervise the motion state of the robot.The experimental part of this paper mainly cotains the underlying communication experiment,the middle layer map construction experiment,the navigation experiment,and the upper layer remote communication experiment.The experimental results achieved the expected goals.