ROS-Based Autonomous Localization And Navigation For Mobile Robots

With the development of computer technology and artificial intelligence,mobile robot has been widely used in the fields of space exploration,industry,military,service,medical and so on.And it can replace the human to complete a variety of work and has broad application prospects.Autonomous localization and navigation technology,which is the basis for robot to complete other more advanced tasks,has important research significance.In this thesis,aiming at the problem of self-localization and navigation for indoor mobile robot,the kinematics model of robot is established,and fusion method of multi-sensors data and improved RRT(Rapidly Exploring Random Tree)algorithm are adopted.The problem of robot positioning and path planning is studied respectively.The main works of the thesis are summarized as follows:1.In view of the shortcomings of existing mobile robot platforms with poor scalability and low code reusability,a mobile robot platform is built up based on ROS(Robot Operating System),which can provide the hardware and software foundation for the realization of algorithms.2.Since the localization of the mobile robot with only encoder data is not able to provide high precision results,an algorithm to fusion encoder and gyroscope data based on Kalman filter is proposed.This algorithm can effectively reduce the influence of angle error on robot localization.3.In view of the problem of error accumulation in relative position of mobile robot,the PROSAC(Progressive sample consensus)algorithm,combined the quality of feature matching with depth information based on RGB-D camera,is adopted to estimate inter-frame movement.Key frames are detected and g2o(general graph optimization)solver is used to make local optimization to get the best pose estimation of key frame.To a certain degree,this method can reduce the influence of cumulative error on the positioning accuracy of robot.4.Considering the influence of nonholonomic constraint of mobile robot on movement efficiency,an improved RRT algorithm is proposed.The algorithm takes the nonholonomic constraint of mobile robot into account in global path planning and it can plan a smoother path,which makes mobile robot more easily get the target position.Finally,conclusions and future work are presented in the end of the thesis.