Motivation

Two of the key principles behind Industry 4.0 (I4.0) are the ability of machines and devices to communicate with one another and the capacity of the resulting cyber-physical system to autonomously perform tasks. But how can this vision be realized? While fine-grained central planning and control clearly has merit in large-scale manufacturing, the approach is not well-suited for highly dynamic and unpredictable environments, where modern versatile collaborative robots (cobots) operate. In such environments, robots must continuously take autonomous decisions based on onboard sensor readings to carry out their task successfully and safely. 

Swarm robotics is an approach to the decentralized coordination of large multirobot systems. In this approach, there is a focus on embodiment, where robots communicate locally with one another, either explicitly or through modification of the environment. By following the core principles of individual autonomy, decentralized coordination, and self-organization, swarm robotics systems can achieve a high degree of scalability, flexibility, and robustness.

Future work

We expect the results of this project to lay the foundation for a new, ground-breaking approach to local coordination in I4.0 that imbues future manufacturing systems with a high degree of autonomy, flexibility, and robustness. We furthermore expect to advance the field of swarm robotics significantly by paving the way for application to more complex robot platforms and cyber-physical systems. Technology providers and manufacturing companies can directly benefit from the technology, since it enables robots to collaborate in a robust and flexible manner.

Current work 

We are applying swarm robotics to I4.0. We envision a future in which centralized planning is limited to high-level decision-making, and where robots use swarm robotics to coordinate and cooperate locally. 

In such scenarios, mobile platforms would, for example, be able to autonomously cooperate with one another to optimize transport of components, they would be able to locally cooperate with robot arms to load and unload payload, and co-located robot arms would autonomously decide when to work together on the same task and when to work in parallel, based on the local situational context.

However, a number of key scientific questions must be addressed before swarm robotics can be widely adopted in I4.0:

• Perception: how to enable robots with limited onboard sensing capabilities to assess the situational context with sufficient accuracy to carry out tasks efficiently and safely in complex I4.0 environments.
• Coordination: how to give robots, potentially with a heterogeneous set of capabilities, the capacity to cooperate on tasks that require multiple steps (e.g. manipulation, transportation, and assembly tasks) in highly dynamic environments.
• Control: how to develop methods for automatic design of decentralized control for robots in an I4.0 context.  
• Validation: How to design and conduct case studies to assess performance.