Design And Implementation Of Vision-based Robot Dynamic Positioning System

Vision-based robot dynamic positioning system has the characteristics of strong programmability,wide adaptability and improved work efficiency and accuracy.It is widely used in 3C,medicine,food,injection molding and other fields.According to the requirements of engineering applications,aiming at vision-based static and dynamic operation problems,a robot visual system is designed and implemented based on the study of visual positioning technology,dynamic control strategy and robot tracking technology.A system solution including robot static grasping and dynamic tracking based on visual positioning is designed.By the study of the relationship between the robot and visual coordinates,a vision calibration model based on perspective transformation and a four-point calibration method are proposed.Then the parameters are optimized by the least square method.Through static workpiece grasping experiments,the static grasping error of robot is within 0.5mm,so the correctness of the visual calibration and positioning method and the grasping precision are verified.For the moving workpieces on the conveyor,the conveyor model is established.A parameter calibration method based on the alignment of the marked points is proposed.Thus the mutual conversion between the coordinate systems of the robot,vision,and conveyor is completed.The dynamic control strategy for real-time updates of conveyors and workpieces is studied.A method is proposed that detects and updates the conveyor status in real time based on encoders.A method is proposed that manages workpieces based on variable-length first-in-first-out queues.As a result,the robot movement is determined based on real-time accurate and non-repeating information of the workpieces.For the problem of tracking different process points with real-time changed position,a method of calibrating the workpiece reference coordinate system and teaching process points is proposed.By mapping the operating point to the real-time workpiece coordinate system,real-time coordinates of any process point on any pose can be calculated.During the dynamic tracking of the robot,the movement process is analyzed in detail and its interpolation points coordinates are calculated.Then to optimize the tracking effect,a trajectory optimization method based on speed compensation is proposed to solve the mechanical impact when the robot motion state is switched.The experimental platform for robot dynamic tracking is set up.Through the experiments to track small workpieces on the conveyor,the contour error of workpiece follow-up is within 1mm when the conveyor moves within 20m/min.The effectiveness and applicability of calibration methods,dynamic control strategy and robot dynamic tracking in this thesis are verified.