RobWoT: Generating Real-Time Digital Twin Simulations for the Robotic Web of Things

摘要

A cornerstone of the Fourth Industrial Revolution is Internet of Things (loT) technologies, which enable unprecedented connectivity in industrial settings, facilitating novel approaches for industrial automation. However, the high fragmentation in the IoT ecosystem hinders the potential of IoT technologies in the industry. To solve this problem, the W3C proposes the Web of Things (WoT) and its Thing Description (TD), a JSON-LD document for describing the API exposed by a device and the inputs and outputs of each interaction. In contrast to simple devices, describing the capabilities of robots and accurately depicting their workspace using the TD is a significant challenge and has to be done manually. Furthermore, there is a lack of simulation frameworks that help validate WoT robots. In this paper, we introduce RobWoT, a method and an open-source simulation framework for automatically generating WoT-compliant Digital Twin (DT) and its corresponding, annotated TD using a Unified Robotic Description Format as an input. These DTs can be used to preemptively test a robot's movement com-mand to prevent malfunctions or collisions, enabling simulation-in-the-Ioop scenarios. We evaluate the execution time of our generation algorithm and the accuracy of a DT's movement compared to a physical robot. Our evaluation shows that our generation method scales linearly with collidable artifacts in the simulation scene and that our DT deviates6.8 mm on average from the target position. Our proposed method makes developing and verifying WoT robotics more standardized, streamlined, and easy to set up.