A Marker And Laser Based SLAM System

With the rapid development of the electronic commerce industry and the express delivery service industry,more and more large-scale warehouses are put into use.Currently in most existing large warehouses,manual picking is usually carried out using trolleys or forklifts.However,workers have limited work efficiency and cannot work continuously with high strength.At the same time,the physical strength of workers is limited.When using forklifts to transport large cargo under fatigue,it is easy to cause accidents bringing accidental losses.These problems have prompted the rapid development of intelligent picking robots.At present,relatively intelligent warehouses use Automated Guided Vehicles(AGVs)to carry goods.Mainly by embedding metal wires or tapes and magnetic points under the AGV running line,then AGV can use magnetic sensors to detect the magnetic field to realize positioning and guiding.This method can only perform the basic handling function,and it needs to change the external environment,which consumes a lot of financial and material resources.Simultaneous Localization and Mapping(SLAM)technology is a vital technology in the field of intelligent mobile robots.It can sense the external environment and construct a map through sensors without changing the external environment.At the same time,it can determine its position based on these data and lay the foundation for the robot’s autonomous movement.This paper details the probabilistic basis in robotics,from Bayesian filters and hidden Markov models,to Gaussian filters and Kalman filters and its extensions.The EKF SLAM algorithm is described in detail below and implemented by the artificial marker Apriltag.In the experiment,the robustness and accuracy of tag identification were tested,and the related algorithms were compared,and the shortcomings of the algorithm were pointed out.In view of the problem of positioning navigation and shelf cargo identification of picking robots in warehouse logistics environment,this paper proposes a SLAM method combining artificial tags and lasers.While the robot was moving,a two-dimensional grid map or a three-dimensional point cloud map containing manual tag information is simultaneously established to ensure the autonomous movement of the picking robot and the identification of the shelf goods.At the same time,the method also includes a loopback detection function,which can correct the accumulated error caused by the laser odometer,thereby obtaining more accurate map information.Finally,the robustness and accuracy of the proposed method are verified by experiments in the teaching building and warehouse environment.