Research On SLAM For In-door Service Robot Based On Binocular Vision

With the development of robot technology,as well as the aging of the world population,more and more service robots have come into families to provide service for people.Simultaneous localization and mapping(Simultaneous localization and mapping,SLAM)problem is the key technology to implement the autonomous navigation and provide service for people of service robots.The vision sensor has many advantages such as low cost,low energy consumption,large amount of information,intuitive effects and so on,so it has become the main sensor to obtain the environmental characteristic information in the SLAM system.In this paper,how to implement the SLAM system by binocular vision is studied and how to design and implement the SLAM system is described.The main research contents include:In this paper,the development of localization and map building technology of service robot is studied,and the advantages and disadvantages of SLAM are analyzed.In order to achieve the binocular vision targeting positioning algorithm,the basic principles and related algorithms in machine vision are studied,including camera calibration,target recognition,stereo matching and 3D information recovery.The calibration method of camera calibration is Zhang Zhengyou calibration method.Using geometric moment invariants combined with Euclidean distance goal recognition algorithm for image matching the shape of each region.Later the matching algorithm based on the weighted average algorithm with arctangent normalized of the three types of regional characteristics was proposed according to the paper requirements with the left image of the target area having been normalized after characterizing subtracting absolute value weighted by regional matching results.Finally,the recovery based on least squares method for three-dimensional information was researched.The SLAM method based on extended Calman filter(EKF-SLAM)is studied in detail.The SLAM system based on binocular vision is established,and its motion model,observation model and data association problem are analyzed.In this paper,the principle of Calman filter is introduced,and the realization process of EKF-SLAM algorithm is explained.SLAM platform for service robot is built.The feasibility of the binocular vision localization algorithm is verified by experiments,and the accuracy of the algorithm is tested.Acquisition of the course by the encoder,binocular vision system to collect the environmentalcharacteristics of the information,the collected experimental data uploaded to the host computer for processing,to verify the feasibility of the SLAM algorithm.