Research On Autonomous Charging Technology Based On ROS System

In recent years,with the continuous maturity of computer technology,the field of work of mobile robots has been increasing.A characteristic of mobile robots is that they need to be charged.In order to enable mobile robots to work for a long time,it is necessary to solve the robot's charging problem.Mobile robots generally build a map of the environment and control the robot to navigate on the map for docking and charging.But this usually has higher requirements for the robot operator and the working environment of the robot,because the traditional method of using lasers and cameras to construct environmental maps does not contain semantic information of objects,and it is not difficult for professionals to operate the map.And because of the defects of the current algorithm,if there are dynamic objects such as people in the process of building the map,these dynamic information will also be added to the map,making the map construction inaccurate.In the end,due to various reasons,the robot cannot be accurately docked with the charging pile.In order to solve these problems in the mobile robot charging process,this paper designs a set of autonomous charging docking system based on ROS(Robot Operating System)system.In order to complete this system,this article did the following research:(1)Researched the use of cameras to obtain environmental information.By transplanting the yolov3-tiny algorithm to the ROS system,the target detection algorithm can more easily communicate with the nodes in the ROS system,and use yolov3-tiny to detect the image,and obtain the position information corresponding to each object in the image And category information.Improved the method of obtaining obstacle data,making the data closer to actual obstacles.(2)Study the fusion of target detection data and depth camera data.Through data fusion of depth camera data and target detection data,semantic information is given to each depth camera data.(3)Researched the construction of environmental map.The construction method of the semantic map is designed.The depth camera data with the object information is used to add the semantic information of the object to the occupied grid map to generate a semantic map.And designed a semantic label display program,so that the semantic map can be displayed in ROS rviz display software.The construction method of static obstacle map is designed to make the map more conducive to robot navigation.(4)The robot navigation and charging pile docking work are studied.It allows the robot to navigate from the starting point to any location on the map that can be passed,and to avoid temporary obstacles.The charging docking method of the robot and the charging pile is designed,and the charging pile is identified through the camera,the positioning of the charging pile is completed,and the accurate docking of the robot and the charging pile is realized.Finally,the experimental data shows that the autonomous charging system of the robot designed in this paper solves the problems of poor readability of the map during charging of the mobile robot,the map contains dynamic obstacle information,inaccurate robot navigation,and low robot docking success rate.The autonomous charging of mobile robots.