Research On Key Technology Of Virtual Teaching Of Industrial Robot Based On Binocular Vision

With the widespread use of robots in the industry,the requirements for robots can not only accomplish simple and repetitive tasks,but also require the ability to apply different environments and can perform complex operations more quickly,accurately and efficiently.How to realize the rapid and accurate teaching of the robot is a hot spot of research.The traditional teaching box is inefficient and is not suitable for the tedious application scene.With the development of sensors,teaching methods such as virtual teaching,force-guided drag teaching,pendant teaching,etc.have appeared.Virtual teaching plays an important role.Virtual teaching is based on virtual reality technology,through human-machine interaction,and users interact with each other through voice,image,force and graph to achieve virtual teaching.General virtual teaching needs to reconstruct the working environment of the robot,such as VR glasses,can also use interactive tools: such as Kinect.In view of the troublesome and time-consuming teaching of traditional robot teaching,in this paper,we propose a binocular vision and augmented reality markers combined teaching system.The system tracks the position and pose of the teaching rod,the teaching rod's position and pose is converted to the robot's base coordinates,to achieve teaching?Teaching rod is composed of ChArUco checkerboard,thimble and so on.The operator can rapidly and accurately teach by simply operating the teaching rod.The main work of this paper is as follows:In order to ensure the accuracy of teaching,This paper uses a binocular camera to measure the position and pose of the checkerboard,and compares the position and pose of the TMC measured by the laser tracker,Inaccurate for the virtual teaching construction environment,in order to ensure the accuracy of teaching,using the binocular vision and the virtual reality ChArUco checkerboard to cooperate,The binocular vision is used to track the position and posture of the teaching rod,and the posture and position is converted to the robot base coordinate system to realize virtual teaching.That is,the position and posture of the teaching rod is equivalent to the position and posture of the end of the machine tool,thereby avoiding the construction of a complicated teaching environment that cannot guarantee the accuracy.The space location,spatial distance measurement and spatial position and pose accuracy were verified by the laser tracker.The absolute positioning accuracy of spatial positioning is less than 3 mm,the accuracy of spatial ranging is less than 0.5 mm,the accuracy of spatial position and pose measurement accuracy around each axis is less than 5 degrees,and translation accuracy is less than 6 mm.When the teaching rod to teach in the operation,it is found that the hand jitter has a great influence on the results.In order to get better teaching effect,the position and posture of the teaching rod of the visual system are reprocessed,and the Kalman filter is used to perform the de-jitter operation.In order to maintain accuracy,processing is only for translational vector processing.We will get the translation vector after removing jitter,and then use the quadratic B spline curve to fit.Finally,the final translational vector and the original pose are used to obtain the final position and pose of the teaching rod.Taking into account the camera's field of vision during operation,it may cause the teaching rod to be occlusion.Proposed anti-occlusion by translating the coordinate system.For the case of partial occlusion in the teaching process,use the ChArUco checkerboard instead of the traditional checkerboard to construct the teaching rod.When the occlusion occurs,anti-blocking is achieved by the translation coordinate system,and the new pose is converted to the original that has been defined in the original coordinate system,and verify the accuracy of pose conversion.The error for each axis rotation is less than 3 degrees.In order to verify the teaching effect,a virtual teaching test platform was built,and the positioning accuracy of the robot was below 3 mm.The accuracy and stability of the spatial positioning by the robot's spatial settled point and the walking space curve verified the continuity of the spatial trajectory teaching,and verified the teaching accuracy of the teaching rod's posture change significantly by teaching a vertical trajectory.The test data shows that the output of the virtual teaching system has good accuracy and stability.