ODIN architecture enabling reconfigurable human – robot based production lines

Abstract

Over the last decades, a significant effort has been allocated by the research community towards improving robot capabilities. These emerged robot cognitive and perception abilities have been driving the swift of production paradigm from rigid automation systems to hybrid and collaborative working environments. Nevertheless, the adoption of robots in lower volume, diverse manufacturing environments is still constrained, due to a) the high integration and deployment complexity of end-to-end integrated robotic applications, b) the limited flexibility of existing solutions in dealing with unexpected events in a dynamic way. This paper presents a modular service-oriented architecture for robotic resources that aims to address those limitations by a) creating digital models, semantically characterizing all production entities and processes in a unified way, b) deploy open interfaces allowing the communication of all robotics hardware and control systems, c) dynamically plan and orchestrate manufacturing operations execution by the involved human and robot resources, and d) collectively share the resources' perceived environment. The overall system includes the integration of task planning, perception, actuation, and robot control modules while Augmented Reality (AR) based interfaces has been considered for including human operators in the loop. The described architecture has been tested and validated in a case study inspired from the automotive industry.