Multi-Agent-Systems in Cyber-Physical Production Systems


  • Birgit Vogel-Heuser
  • Marga Marcos
  • Kira Barton
  • Haitham Elfahaam
  • Luis Ribeiro
  • Paulo Leitão
  • Armando Walter Colombo
  • Arndt Lüder


  • Birgit Vogel-Heuser, Technical University of Munich, Germany (L)
  • Kira Barton, University of Michigan, Ann Arbor, USA
  • Armando Colombo, HS Emden-Leer, Germany (L)
  • Friedrich Durand, Afag Holding AG, Switzerland
  • Cesare Fantuzzi, University of Modena and Reggio Emilia, Italy
  • Christoph Hanisch, Germany
  • Max Hoffmann, RWTH Aachen University, Germany
  • Marga Marcos, University of the Basque County, Bilbao, Spain
  • Astrid Nieße, Leibniz University Hanover, Germany


In the era of industry 4.0, the production system is not only referred to as an automated production system, but as a more complex, interconnected, intelligent socio-technical system. Due to the increasing automation and thus, increased complexity, agent systems are a flexible and convenient way to implement in production plants. However, the development also comes with challenges. Some of those challenges and major topics of multi-agent-systems (MAS) are addressed in this workshop. We start with talks from academia and industry, including the reasons to change to agent oriented concepts, the challenges to be addressed, like: agent interfaces, -patterns, -metrics to measure the autonomy and benefit of agents, how to embed artificial intelligence. We discuss how to apply agents in Industry 4.0 and Engineering.
The goal of this workshop is to provide a broad coverage and dissemination of foundational research in MAS among researchers, academics, and practitioners. This workshop will highlight current topics related to the challenges of MAS in automation science and engineering, and should include keynotes, group work that encourage the participation of attendees in active discussion. The workshop will be used to further evolve the future roadmap for MAS in CPPS. The following topics will be considered:
• Introduction on multi-agent-systems in physical production systems
• Interfaces between MAS and other systems, and between MAS and field-level control
• Machine Learning and autonomous systems in Industry 4.0
• Patterns to realize MAS
• Metrics for autonomy of MAS, benefit of MAS
• Engineering of MAS and autonomous systems
• Social-technical industrial agent-based systems
• Design and Engineering EthicsSocietal Implications of agent-based CPPS
• Human and Agents realising autonomous systems
• Human in the Loop

The workshop is supported by TC 3.1, GMA FA 5.15 and IEEE TC on Industrial agents. Speakers marked with an (L) are lecturers on the annual International Summer School on Industrial Agents (2019 in Bragança, Portugal; 2020 in Warwick, UK).


Challenges in Industrial Agent-based Cyber-Physical Production Systems, Birgit Vogel-Heuser

Industrial agent concepts and technologies can be used to address the needs of Cyber-Physical Production Systems (CPPS). To achieve this, their interplay, especially in industrial settings, needs to be better interpreted and assessed. In this opening talk, key directions for future investigation derived in prior workshops will be introduced, such as the design patterns for multi-agent systems, interoperability and interfaces between the agents and the CPPS architectures, metrics to evaluate the quality of an agent-based CPPS, and Artificial Intelligence enabled by agent implementations.

Two reasons for change, Friedrich Durand

The automated industrial world opens up new opportunities through the effective use of digital manufacturing, robotics and automation. For many manufacturers, however, Industry 4.0 raises the sobering question: If digital manufacturing and automation are the future, how should these technologies look in their companies? How could autonomous systems provide a good solution against the interdependencies? Answers will be given in this contribution.

Morning break

Agents for cloud computing in industrial applications, Cesare Fantuzzi

This contribution aims to describe architectures and algorithms to develop a cloud-based system for data management and processing for industrial plants spread in wide geographical areas. The contribution will discuss in particular (a) distribution of agents between edge and cloud, (2) algorithms for data streaming processing, (3) description of a real case study: predictive maintenance of the fleet of autonomous vehicles.

Agents in Industry 4.0, Armando Colombo and Christoph Hanisch

The growing complexity of production systems requires appropriate control architectures that allow flexible adaptation during their runtime. In the context of industry 4.0, a number of terms exist that are of importance for achieving this. Thus it leads to the need to classify the terms of cyber-physical systems, physical agents or holons, the RAMI 4.0 reference architecture as well as their relationships.

Product Agents for advance manufacturing, Kira Barton

Application of Multi-Agent Systems for production system control has high  potential to change systems engineering for manufacturing in general. This is based on two main facts: (1) agents enable new control architectures with advanced production system functionalities related to flexibility and adaptability, and (2) agents also enable the integration of intelligence across the manufacturing enterprise, from the products to human collaborators. Challenges and opportunities related to this integration will be discussed.

Agents for data-driven approaches for production planning and control, Max Hoffmann (tentative)

Challenges and potential benefit of MAS, All the participants in small groups, Moderators: Birgit Vogel-Heuser, Kira Barton, Max Hoffmann

Conclusion: Presentation of group work results, All the participants, Moderators: Birgit Vogel-Heuser, Kira Barton

Lunch time

MAS from production systems to cyber-physical energy systems: new challenges, Astrid Nieße

One of the major challenges of designing and implementing cyber-physical energy system lies in the development and testing of technologies and concepts to guarantee functionality, stability and safety during operation. Cyber-physical energy system unite formerly separate domains from the physical energy systems, e.g., electric power grids, distributed energy resources. Which specific additional challenges are to be addressed for such systems compared to production systems.

Deep dive and action plan on selected aspects from small group sessions in the morning, All the participants in small groups, Moderators: Armando Colombo, Marga Marcos

Conclusion, remarks, and the roadmap for MAS to enhance CPPS and CPES, Moderators: Armando Colombo, Marga Marcos, Birgit Vogel-Heuser, Friedrich Durand