Research And System Implementation Of Autonomous Navigation For Indoor Mobile Robots

With the development of technology,intelligent robots have almost entered all aspects of human life,and widespread intelligent robots are changing people's lives.Facing the ever-expanding market of service robots,robot navigation technology which is one of the core technologies in the field of robotics,has received more and more attention.This thesis focuses on the navigation technology of indoor mobile robots.The main contents are as follows.Firstly,the status situation of mobile robot navigation technology in domestic and foreign is investigated,and the key technologies of navigation are summarized.The mobile robot platform used in this thesis is introduced.The framework of robot system is designed,and the hardware platform and ROS software system are built.The kinematics modeling of the three-wheel omnidirectional wheel robot in this paper is carried out,and the test datas of the wheel odometer is obtained.Secondly,the environment color image and the depth image are acquired by the depth camera.The feature points are extracted according to the Features from Accelerated Segment Test(FAST)method.The motion relationship between adjacent image frames is estimated,and so the visual odometer test data can be obtained.The Extended Kalman Filtering(EKF)method is used to fuse the wheel odometer,Inertial Measurement Unit(IMU)and visual odometer data to reduce the cumulative error and drift error of the individual odometer;,a two-dimensional grid map of the indoor environment is constructed by using the fused odometer data,and Simultaneous Localization and Mapping(SLAM)function is implemented.Thirdly,the global and local path planning methods of mobile robots are studied.The navigation system based on grid map is built to implement the function of fixed-point navigation of mobile robots.The stability and reliability of the path planning algorithm are verified by the mobile robot experimental platform.Fourthly,we design a charge docking system and study the close-range docking algorithm of infrared and ultrasonic fusion.At the same time,combined with the fixed-point navigation method above,the docking of the mobile robot from any positions is realized.In addition,based on the theory of finite state machine,in the ROS platform,the task script is constructed by using the SMACH library.And the task scheduling of continuous navigation and automatic recharge is realized through the transition between states.Finally,the task experiment is carried out in the indoor environment,which verifies the reliability of the composition and navigation algorithm.Implementing the specific navigation task scheduling through state machine,and accomplish automatic recharge when the power is low.