An Switchable Multi-resolution Architecture of Cyber-Physical Manufacturing Systems (CPMS) for Industrial Robots Collaboration

Abstract

For complex manufacturing systems, fast, accurate, and reliable modeling and simulation of the real world, as well as the interaction from the simulation world to the real world, is required. The development of Cyber-Physical Systems (CPS) and Internet of Things (IoT) enable real-world manufacturing systems and their cyber world to form a manufacturing-oriented Cyber-Physical Systems - Cyber-Physical Manufacturing Systems (CPMS). However, the low performance of the edge, and the heavy storage and computing burden of the cloud, cannot meet the fast and accurate requirements of CPMS. To address these issues, this paper proposes an cloud-edge collaboration architecture of Cyber-Physical Manufacturing Systems intended for industrial robots collaboration. In the architecture, the key planning and decision are placed at a central computing station and the trivial calculation tasks are placed at the information shell of the manufacturing equipment. Specifically, robotic arms, AGVs and other manufacturing nodes are designed to store and perceive the environment and self-state, run with basic kinematics and kinetics. Reconfigurable computing nodes based on FPGA performs trivial logical calculation tasks. The manufacturing could is designed to plan and control all holonic nodes based on multi-agent deep reinforcement learning. The collaboration between robotic arm and AGV is studied as a case. The solution based on the proposed framework is given for the issue. The feasibility of the framework is verified by simulation and derivation.