Architecture for Industry 4.0-based Manufacturing Systems

Abstract

Industry 4.0 or the fourth industrial revolution refers to the trend of integrating cyber-physical systems into manufacturing. Outfitting manufacturing units with an array of sensors/end effectors provides the ability to build a virtual model of the unit. This virtual model can, in turn, be used for automated decision making leading to improvement in performance. We propose a next-generation computational framework for dynamic, data-driven optimisation of production. In keeping with the principles of Industry 4.0, our architecture is easy to reconfigure and connect to various hardware/software devices allowing for quick reconfiguration of factories. It has a live representation of the real world enabling managers and other users to keep track of activities in the factory. It also provides users with statistics and other abstraction tools to augment decision-making capabilities and has the ability to automatically allocate resources thereby allowing for nearly autonomous operation of the factory. We further demonstrate an implementation of the proposed architecture on a model problem developed in conjunction with Foxconn. The implementation consists of a Robot Operating System (ROS) based event simulator, ROS based visualization platform and multiple planners. We implement a baseline strategy inspired by the working of Foxconn factories and then show how our architecture improves upon the performance of the baseline method. The proposed architecture yields an almost 100% improvement over the baseline strategy.