Research On SLAM Algorithm For Mobile Robot Based On Binocular Vision

At the same time,positioning and map building(SLAM)is the most advanced technical problem in the field of mobile robot research.Robot autonomous navigation is the future development trend,SLAM robot is the basis of navigation capability.With the image processing technology matures,the visual sensor in the mobile robot SLAM research is widely used.It is based on the advantages of low cost,low energy consumption,large amount of information collected and intuitionistic,which has been identified as one of the most reliable sensors of robot SLAM.This paper focuses on the application of binocular vision robot SLAM algorithm.This paper is divided into three parts.The first part analyzes the core difficulty of the mobile robot SLAM system,and establishes the robot motion model,the observation model,the odometer model and the camera model,and calibrates the camera model.The establishment of the robot model has laid a theoretical foundation for the experimental system platform.The second part focuses on the ORB feature extraction and matching algorithm in binocular stereoscopic vision,and proposes an optimized algorithm for false matching.The algorithm is used to eliminate the mismatch points in the image matching process,and to estimate the single matrix by limiting the number of iterations and to judge the interior points again.Experiments show that this method can effectively eliminate the error matching and improve the matching accuracy.The third part studies the basic principle of FastSLAM algorithm and optimizes the FastSLAM algorithm.The original FastSLAM algorithm is located through a particle filter to implement the map creation problem with N independent extended Kalman filters.Optimization of the algorithm the new pose process also uses the particle filter algorithm,but in the N independent EKF to achieve the map to create the problem on the use of UKF algorithm for posterior probability approximation to complete the road sign estimation.The feasibility of the optimization algorithm is verified by experiments.At the end of this paper,the combination of theory and practice,design and build a mobile robot SLAM experimental platform.This paper chooses the four-wheeled robot as the main body of the mobile robot,and establishes the binocular vision ranging system to realize the collection of the environmental characteristic information.The platform has the ability to collect information about the surrounding environment,but also have the ability to move,pose detection and calculation.Binocular vision system will be collected information sent to the host computer for processing,and ultimately complete the binocular mobile robot SLAM task.From the experimental results of the platform,the feasibility of the optimization algorithm is verified.