Research On Teleteaching System Of Welding Robot Based On Virtual Reality

Currently high-quality welds often can only be completed by skilled welders whose torch-manipulating skill is believed playing a critical role.If those irregular and unpredictable torch-maniplating movements of skilled welders can be recorded and taught the robot arm to replay them,the robotic arc welding quality may be improved,and this is also the fundamental for developing intelligent robot who possesses similar torch-maniplating intelligence as a human-welder has.In view of above background and requirement,a welding robot tele-teaching system based on virtual reality(referred to as "teaching system" below)has been designed and developed in this dissertation.Virtual reality technology has good immersive experience.Using virtual reality technology can effectively record the welder's torch-manipulating movements and transferred to real robot movements through the human-robot interaction.The overall frame design is first conducted,including the system definition,requirements analysis and structural design.The teaching system consists of two independent parts: the virtual reality operating side and the real robot executing side,The VR side is developed with HTC VIVE VR kits and Unity3 D engine.The real robot side is developed with YASKAWA MH6 robot and MOTOPLUS environment.The architecture of the teaching system is designed,and the specific functions of the main modules are defined.Ethernet is chosen as the communication interface between the VR operator and robot executor,and the communication scheme is designed by mixed using of TCP protocol and UDP protocol.The specific modular and functions are then further developed and realized according to the mainframe design.Standardization and interfacing method are used during the coding process,including the VR scene building,user interface design,joint hinge reconstruction design,virtual robot simulation module design,operation module design,coordinates conversion module design,step-by-step teaching module design,etc.The forward and inverse kinematics method and pose representation transform method are mathematically resolved and realized in the software.The controlling software of the real robot side is developed via API provided by MOTOPLUS.The torch movement is simulated and controlled by a handle controller associated with the VR system.The movements of the handle carried by the human operator is recorded and transferred to virtual robot movement and taught to the real robot.The essential issue of the control strategy is the mapping relationship between the movement of the real robot torch and the handle controller during the human-robot interaction.A satisfactory control strategy can bring more intuitive and flexible operation experience and more reliable and stable operation effect.Four kinds of control strategies,which are named "Mastery" Strategy,"Gesture" Strategy,"Rocker-like" Strategy,and "Thrust" Strategy are implemented and analyzed.The results show that the "Gesture" strategy is the optimal strategy for the operator to use the trunk core muscle group to exert force for obtaining better stability.The "Gesture" strategy have best adaptability to several trajectories,and is thus the optimal control strategy for teaching system.Finally,the functionality and performance test of the teaching system are carried out.The relationship model between the time delay and position(angle)delay with line(angle)speed is established.The test results show that when the speed is lower than 40 mm/s or 40°/s,the time delay of the teaching system is less than 0.12 s and the trajectory error is less than 0.06 mm/0.1 degree.The overall operation performance fully meets the requirements of welding torch manipulating.The dynamic response of the VR robot is higher than that of the real robot executing side,due to the lack of openness of the controller of the real robot.