Research On Virtual And Real Synchronous Control Of Mobile Robot Arm Based On Digital Twins

Aiming at the problems of single control method,less intuitive operation and low intelligence of automatic equipment of oil rig with mechanical arm,digital twin technology was introduced,and a six degrees of freedom tracked mobile manipulator was taken as the research object to realize the virtual and real synchronization control of mobile manipulator.At the same time,the Near End Strategy Optimization(PPO)algorithm was introduced,and the Unity Machine Learning Agent(ML Agent)framework was used to achieve synchronous tracking of target objects at the end of virtual and real mobile robotic arms.The main research work and achievements include:(1)Constructed a virtual and real synchronous control system architecture for digital twin mobile robotic arms.By determining the control strategy of the digital twin mobile robot arm and the five dimensional model structure of digital twin mobile robotic arms,and used Unity3D as the digital twin development platform,a virtual and real synchronization control system framework for digital twin mobile robotic arms was established.(2)Built a digital twin simulation environment for mobile robotic arms.Through the collaborative work of SolidWorks,3Ds Max,and Unity3D software,a digital twin model of a physical mobile robotic arm and drilling equipment was constructed.A digital twin simulation scene of a mobile robotic arm composed of a mobile robotic arm digital twin model,virtual drilling equipment,and the surrounding environment of a virtual well site was constructed in Unity3D.(3)Designed a virtual and real synchronization and target tracking control system for mobile robotic arms.Used serial communication protocol to establish virtual and real bidirectional connections,four human-machine interaction interfaces that can switch between each other were designed using various UGUI components,achieving bidirectional synchronous control of virtual and real mobile robotic arms in two modes: model driven and virtual and real fusion.By introducing the Near End Strategy Optimization(PPO-clip)algorithm with clipping objectives,a virtual simulation training environment for mobile robotic arms was constructed.The task of tracking virtual target objects at the end of the digital twin model of a mobile robotic arm was trained through two methods: multi-agent training scenarios and simplified multi-agent training scenarios,achieving the control of autonomous synchronous tracking of virtual and real robotic arms.The actual test results of the system indicate that the use of digital twin technology to achieve virtual and real synchronization control of mobile robotic arms has achieved the expected effect,and deep reinforcement learning algorithm has achieved fast and accurate synchronization tracking of the target at the end of the virtual and real mobile robotic arm.This research lays a foundation for the research of digital twin technology of oil rig,and provides a new way to further promote the digital and intelligent process of oil rig.