Multisource Model-Driven Digital Twin System of Robotic Assembly

Abstract

The digital twin technology effectively fuels improvements in the quality and efficiency of robotic assembly, especially for sophisticated processes. High-quality digitalization requires a comprehensive description and real-time rendering of system, which are challenging. This article aims to demonstrate a novel multisource model-driven digital twin system, which is based on the geometric, physics, and sequential rule descriptions for producing a precise and real-time simulation of a robotic assembly system. In our system, the digital counterpart rendered with virtual reality technology is created and updated by the real environment information. A 3-D graphic model is reconstructed first using the geometric information of surroundings captured by the depth sensor. During interaction, a fast and efficient approach is proposed to generate the contact force and render the object deformation in the virtual environment. The virtual space also provides the sequential rule that is based on the danger field, which helps to constrain the operation of digital twin. At last, the experimental platform is established in which the virtual space is at an update rate of 100 Hz, while the automatic sorting task is performed and verified; the results show the effectiveness and applicability of the method to digital twin system.