Quasi-Decentralized Cyber-Physical Fabrication Systems—A Practical Overview

Building an effective cyber-physical system is difficult due to the overall complexity of technologies on their own, challenges with the interaction between all parts of workflow and applicability issues. This makes the real-world application of complex cyber-physical systems only available to the big industry parties or high-tech startups, leaving small and medium-sized businesses behind. Therefore, the democratization of such applications is a reasonable goal to achieve. Cyber-physical systems of this kind are in the experimental stages and incorporate robotics, IoT, materials science, visual and 3D-scanning techniques, and machine learning (ML) tools all under the heading of Industry 4.0. This paper covers specific practical approaches in several application fields using robotics and IoT. We use custom-built hardware and software setups together with standard frameworks and the MQTT protocol for different applications. Due to the practice-driven approach, the paper will illustrate both positive and negative effects. We describe the use of our systems in several case studies: Multi-layer automated robotic concrete spraying using ML, IoT spatial awareness via sensors (such as Lidar, Kinect), robotic multi-axis milling and quasi-autonomous robot movements. A unifying issue is a decentralized approach of modular IoT elements that we grouped to achieve specific tasks. The paper illustrates how these elements exchange data and communicate, and all services are controlled on each computer instances, connected to an IoT network to ensure a high level of system stability.