Research On Automatic Hand-eye Calibration System For Bin-picking

At present,robot Pick & Place applications based on planar two-dimensional machine vision guiding have been quite mature.While Bin-picking applications based on three-dimensional machine vision guiding,which can recognize and pick targets scattered in three-dimensional space,are still rare.It is foreseeable that research and breakthrough on Bin-picking system will be of great significance in improving the automation and intelligent level of robots and expanding their application fields.Bin-picking system is a hand-eye cooperative system consisting of a vision guidance system and a robot.The purpose of hand eye calibration falls in determining the relative position and pose between the vision system and the robot end effector.Since the target's position and pose acquired is relative to the camera coordinate frame,it has to be converted to the tool coordinate frame at the robot end.Hand-eye calibration is the prerequisite to accomplish the conversion.Aiming to realize a Bin-picking system,this paper focuses on research of hand-eye automatic calibration technology of machine-vision guidance.The main work of this paper is as follows:1)Based on study of the traditional camera and hand-eye calibration algorithm,according to conditions that the unique solution of the hand-eye calibration equation exists and factors to reduce error sensitivity and to improve calibration accuracy,process of calculating and optimizing coefficient matrix of the hand-eye calibration equation is given;camera calibration and hand-eye calibration algorithm is implemented using OpenCV,the open-source computer vision library.2)Design of Bin-picking hand-eye system.Eye-in-Hand model is adopted.Functionally the system consists of mainly the robot subsystem and the machine-vision guidance subsystem.Design of the robot subsystem mainly involves selection of the robot,control and programming method of the robot.Design of the vision guidance subsystem includes hardware selection and the design of each software function module.The integrated design of the robot and the vision system is the focus of the design of the automated hand-eye calibration system.The main content is to design the communication format and robot program flow of the robot and the visual system.The robot is programmed according to the specified routine to achieve synchronization collaboration with the vision-guidance system;a unified definition of the meaning of robot movement task parameters was given.3)Automatic calibration of the hand-eye system has been done.The intrinsic and extrinsic parameters of the camera are obtained as the input of the subsequent hand-eye calibration experiment.The hand-eye calibration module obtains motion parameters from the robot through the communication interface and automatically calculates and selects optimal coefficient matirix according to the algorithm.Re-projection was used to evaluate the precision of calibration results.Results show that the system designed and implemented is able to accomplish the automatic calibration of the hand-eye system with high precision.4)Overall Bin-picking experiment is carried out.Results show that the system designed and implemented is able to accomplish position and pose estimation,coordinate transformation and final automatic picking of randomly scattered targets.The accuracy of the hand-eye calibration outcome and the effectiveness of the system design is then verified.