Research On Indoor Localization Technology For Mobile Robot Based On Passive Beacon

Because the self-localization is the premise for the mobile robot to accomplish different complex tasks,the positioning technology is more and more popular in the field of mobile robot.In recent years,with the rapid development of robot technology and the growing market demand for industrial and service robots,the positioning technology of mobile robot has attracted wide attention from the academic and industrial fields,and attracts a lot of resources for deeper theoretical research and application exploration.However,the problem of accuracy,real-time and adaptability in complex scenes are still not well solved at present.These problems have led to the limitation of the application of mobile robots in many situations.Therefore,in this paper,a real-time high-precision indoor positioning technology for mobile robots Based on passive beacon is studied by using visual and 3D lidar sensors.The main contents are as follows:(1)A kind of artificial markers that can be accurately identified in poor lighting conditions or blurred images is designed.By using the parity check principle of Hamming code,the coding mechanism of the artificial marker is proposed,so that the marker can conduct self-verification,judge the validity of its own ID,and greatly improve the reliability of the marker recognition.In addition,the appearance design of the marker is also introduced in this chapter,as well as the generation and identification process.(2)A visual positioning system based on artificial marker is introduced.Graph-based optimization technique is used in this system.The position of each marker in the global image coordinate system is taken as the node of the graph,and the relative position between each two markers is used as the edge of the graph.By analyzing the camera distortion model and 3σ principle,a camera uncertainty model is proposed,which can be used to describe and update each edge in the graph through Bayes estimation.In this way,the marker map can be more accuracy after the graph optimization.Then,the robot can be located according to the marker positions in both the marker map and the current frame image.(3)An improved algorithm of binocular odometer is proposed and closed-loop correction is performed by the detection of artificial marker.In the front end of the algorithm,the stereo images are captured using a binocular camera and sub-pixel-level matching point pairs are obtained through binocular matching algorithm.And the 3D information of these feature points is calculated in the camera coordinate system.Then the pose transformation is acquired using RansacPnP algorithm based on the 3D-2D matching point pairs.And the transformation is used as the initial value to do the local optimization by minimizing re-projection errors.Additionally,an uncertainty model is proposed based on the Hamming distance between the descriptors of matching point pairs.The model provides the information matrix for the constraints in the local optimization,which can improve the positioning accuracy.In the back end of the algorithm,reliable closed-loop information,artificial landmark,is detected using the monocular camera shooting vertically upwards,and then the global pose optimization is conducted.According to the motion model and working scene of the AGV,an optimization method based on global plane constraint is proposed to reduce the error of localization system.(4)An odometry estimation method based on 3D laser radar is proposed.The algorithm overcomes two big pain points of Velodyne data,sparsity and discreteness,in a fast and efficient way,and achieves accurate pose estimation.First,the ground points are separated through an angle based filter.And the rest scan points are classed through their covariance matrixes generated through each point and its several closest points.Through this way,the non-ground point cloud can be separated into ground point clouds,planar point clouds,line point clouds and unorganized point clouds.When implementing the odometry estimation,it’s much more efficient and accurate to register the sub-point clouds of the same kind.The local map is built with the last several frames to overcome the discretization problem.Then,frame-to-local map matching,instead of frame-to-frame ICP strategy,is used to implement the odometry estimation.