Design And Implementation Of The Security And Inspection Robot's Navigation System

In this ever-changing era,people's lives have gotten more and more abundant because of the development of science and technology.In recent years,the field of intelligence has attracted increasing attention from everyone,and intelligent robots,as smart high-end technology equipment,are also popular.Because the application of intelligent robots is used very widely,and security is an issue that modern people pay more and more attention to,the security and inspection robot produced by the combination of intelligent robot and security concept naturally becomes an important direction for robot research.As the core part of robot autonomous navigation,navigation system is an essential part of robot research.The research content of this dissertation is mainly based on the security inspection robot project.The navigation system of the security and inspection robot is designed and implemented around this project.To achieve this goal,the main work of this paper is as follows:1.The paper will analyze the requirements and function of the navigation system according to the actual use scene of the security inspection robot.On this basis,the overall logical framework and structure of the navigation system will also be designed.2.For the detailed design and implementation of the image perception module in the navigation system,an image calibration algorithm is proposed and it is used to perform image preprocessing.Based on this,an image segmentation algorithm based on deep learning is designed,including the selection of input data and the design of network structure.The generation of training data sets and the extraction of semantic tags are completed at the same time.Finally,the network is used to complete training and testing,and experimental results are obtained and analyzed.3.Through the study of global path planning,the dissertation firstly proposes a method based on GPS point-based topological maps in the modeling of environmental maps,and then uses the least squares method to fit the routes.Meanwhile,the Dijkstra algorithm of path search is improved with the consideration of situation of the intermediate node set.In addition,the strategy of re-planning the path is proposed when the robot deviates from the route,and a back-end management system is implemented on the server,which includes the creation of a map and a path plan.4.According to the research of close-range planning,the dynamic window method is improved for partial path planning and obstacle avoidance.In addition,it must be taken into account to propose a simple and efficient algorithm,combined with an improved dynamic window method,when the partial path planning is error and the robot cannot reach the specified position,in order to achieve the smooth navigation of the security inspection robot to in both normal and abnormal conditions.Via the implementation of the above-mentioned key navigation technologies,the integration of various modules of the navigation system,and the examination of the various functions of the system after the design and implementation,the experimental results prove the feasibility and practicality of the navigation system of the security inspection robot in the actual scene.