Research On Teaching Programming Technology Of Industrial Robot Based On Augmented Reality

Robot programming technology is a hot issue in the field of robot research.The development of its programming technology plays an important role in the promotion of intelligent manufacturing technology applications.The current mainstream robot programming methods include online teaching programming and offline programming.The robot programming is cumbersome and requires high professionalism from programmers,which limits the wide application of robots in various industries.Exploring and improving existing robot programming methods,lowering the threshold of its programming technology,and improving the intuitiveness of robot programming are urgently needed issues.This paper proposes an industrial robot teaching programming method based on Augmented Reality(AR)for six-degree-of-freedom industrial robots(hereinafter referred to as robot AR programming).This method uses smart terminal devices(such as mobile phones and tablet computers)that support augmented reality functions as the AR handheld input terminal of the robot's posture.The user records the movement trajectory and posture of the terminal in the virtual space through the App program in the terminal,and uses the BP neural network to predict and compensate the motion error(hereinafter referred to as AR error)generated when using the handheld input terminal plans the robot trajectory,and the compensated trajectory data is driven by the joint data calculated by the robot forward and reverse kinematics models to drive the virtual robot to move,then realize robot AR programming.The main research work of this article is as follows:(1)The robot AR programming system scheme is elaborated in detail,and the causes of AR errors in the process of the robot AR programming system are analyzed;taking the HSR-JR605 industrial robot as the research Object,establish its digital prototype model,use D-H method to build its kinematics model and solve the robot's positive and negativesolutions.(2)Research and design the data collection and processing scheme in the robot AR programming system.An AR error model training data collection platform was built,and the collected data was subjected to abnormal numerical processing through the3 ? principles.The robot data collection process through the robot secondary development API was not synchronized with the AR handheld input terminal for data collection.The problem is not equal to the collection frequency,and the data collected by the AR handheld input terminal and the robot is re-collected at equal time intervals;finally,the data is smoothed by sliding smoothing,SG filtering,and first-order exponential sliding smoothing..(3)The AR error prediction model is constructed using a three-layer BP neural network.First,the coordinate conversion relationship between the virtual space coordinate system and the real robot Cartesian right-handed coordinate system is analyzed,and then the AR error prediction model is built using BP neural network,which can effectively reduce the AR error.Finally,the trajectory data after AR error compensation is smoothed by the cubic spline smoothing method.(4)Developed a set of software with functions of data acquisition,robot kinematics simulation,AR error prediction and compensation.Finally,using HSR-JR605 and i Phone XR as the experimental objects,the robot AR programming is performed on a plane curve.After the error compensation is performed through the BP neural network,the AR error of the AR handheld input terminal in the x and y directions is reduced by 37% on average,it can effectively reduce the AR error.