Research And Development Of Prison Robot Mapping And Positioning Algorithm Based On Laser Radar And Vision

Along with the progress of science and economic development,the national demand for quality of life has been continuously improved.The traditional security industry has begun to develop towards automation and intelligence,security robots came into being.Compared with traditional security,relying on a large number of sensors,manpower and material resources,security robots work more efficiently and more flexible,and it can completely avoid work abnormalities caused by human emotions and physical factors,correspondingly,due to the importance of its mission,security robots often have high requirements for the accuracy and robustness of their intelligent algorithms.As a special application of security robots,the main purpose of the prision robot is to replace the prison guards to complete the night high-intensity inspection work,require robots to patrol the prision without supervision and provide feedback on the status of prisoners in each escort.All these tasks have high requirements for the positioning algorithm of the robot.In view of this,this paper will study the mapping and positioning algorithm of the prision robot.The main content of this paper is as follows:1.Briefly describe the main functions of the prison robot,emphasizes the important role of the mapping and positioning system in the entire robot system,and introduce its main hardware components and the software architecture of the mapping and positioning system.3.Analyze the principle and implementation process of particle filter based SLAM(Simultaneous Localization and Mapping)algorithm,descripe the initialization process,sampling,scan matching,weight calculation and other steps' calculation process and results.And aiming at the high requirement of odometer accuracy of this algorithm,a SLAM algorithm based on least squares matching is proposed.Algorithm works in two threads,estimating the probability value of the map at the laser endpoint using the bilinear difference method,calculating the relative transformation of robots in a local map by constructing a least squares error function,and use this transformation value to extend the particle set to complete the entire algorithm.3.Aiming at the environment of specific prison with few characteristics and high positioning accuracy,a positioning algorithm based on visual image processing is proposed.The algorithm is divided into two steps,complete the rectangle of the window extraction by extracting the image edge and comparing the image similarity,and combine the camera imaging model to complete the first remote calibration.Through preprocessing the close-up image,the straight line extraction of the railing completes the second close-range calibration,ensuring that the robot can accurately dock at the mission location.4.Verify the accuracy of the SLAM algorithm and the visual image localization algorithm by simulation experiment,laboratory environmental experiment and prision environmental experiment.