Holonic manufacturing

Holonic manufacturing is a manufacturing control paradigm that organises production systems as networks of self-regulating modules called holons. Each holon is simultaneously autonomous—capable of independent decision-making—and cooperative, acting as part of a larger productive whole. The approach draws on Arthur Koestler's philosophical concept of the holon and was formalised through the international Intelligent Manufacturing Systems (IMS) research programme in the 1990s as the Holonic Manufacturing Systems (HMS) initiative.

Holonic manufacturing attempts to address a longstanding tension in production control between centralised and decentralised planning. Centralised planning is generally efficient, but vulnerable to disruption, where decentralised planning is less vulnerable to disruption, but potentially less efficient. By organising production resources, orders, and product knowledge into cooperative holons arranged in a holarchy, the paradigm attempts to create systems that can reconfigure themselves in response to machine failures, new orders, or changing product requirements without requiring central coordination.

Background: the holon concept

Arthur Koestler introduced the term holon in The Ghost in the Machine (1967), combining the Greek holos (whole) with the suffix -on (suggesting a particle or part, as in proton).^[1] A holon exhibits two simultaneous tendencies:

Self-assertive tendency: the holon functions as an autonomous whole, governed by its own rules.
Integrative tendency: the holon participates as a subordinate part of a larger system.

Koestler described this duality using the metaphor of Janus, the two-faced Roman god. A holarchy is a hierarchy of holons in which each level is both a whole relative to its components and a part relative to the level above it. Koestler distinguished holarchies from conventional hierarchies, in which each element is purely subordinate, and from heterarchies, in which elements are purely equal.

Koestler presented a more formal treatment of holarchies in his appendix "General Properties of Open Hierarchical Systems" and at the Alpbach Symposium (1968).^[2]

History

IMS feasibility study and the HMS Test Case (1992–1994)

In February 1992, the governments of the United States, Japan, the European Community, the European Free Trade Association, Australia, and Canada initiated the Intelligent Manufacturing Systems (IMS) international research programme with a two-year feasibility study.^[3] Among the six international consortia selected for one-year test cases was the HMS Test Case, formally titled "Holonic Manufacturing Systems: System Components of Autonomous Modules and their Distributed Control."

The HMS Test Case consortium comprised 31 partners across five continents, including industrial participants (such as Rockwell Automation/Allen-Bradley, Hitachi, BHP, and Alcatel), universities (including KU Leuven, Queen's University, and the University of Calgary), and research institutes. Coordinating industrial partners were BHP Co. Ltd. (Australia), Queen's University (Canada), Softing GmbH (Europe), Hitachi Ltd. (Japan), and Rockwell Automation/Allen-Bradley (USA).^[3]

James H. Christensen is credited with translating Koestler's philosophical concept into manufacturing engineering terms, presenting an initial HMS architecture and standards directions at the First European Conference on Holonic Manufacturing Systems in Hannover in December 1994.^[4]

Full HMS Project (1996–2006)

Following the feasibility study, the full HMS Project commenced in 1996 and ran for approximately ten years, producing reference architectures, software frameworks, and industrial demonstrators. The project established a common vocabulary and architectural principles that subsequent research refined and extended.^[5]

Core concepts

Definition of a manufacturing holon

The HMS consortium defined a manufacturing holon as an autonomous and cooperative building block of a manufacturing system.^[6] It possesses:

A control part, responsible for decision-making and communication.
A physical part (optional), such as a machine, robot, or transport vehicle.
An information part, containing the knowledge needed to fulfil its role.

Holarchy and negotiation

Manufacturing holons are organised into a holarchy. Unlike a rigid hierarchy, a holarchic structure can dynamically reconfigure—a resource holon can join or leave a production cluster as conditions change. Communication between holons typically uses negotiation protocols, often modelled on the contract net protocol: an order holon announces a task, resource holons submit bids, and the order holon awards the task to the best bidder.^[7]

PROSA reference architecture

PROSA (Product–Resource–Order–Staff Architecture) is a holonic manufacturing reference architecture developed at the Production, Machine Design and Automation (PMA) division of KU Leuven by Hendrik Van Brussel, Jo Wyns, Paul Valckenaers, Luc Bongaerts, and Patrick Peeters. The foundational paper has accumulated over 1,400 citations.^[6]

PROSA defines four types of holon:

Product holon: Holds the complete lifecycle information for a product type, including user requirements, engineering design, process plans, bill of materials, and quality procedures. The product holon contains non-deterministic process plans that represent the range of valid manufacturing sequences.

Resource holon: Represents a production resource—a machine, robot arm, conveyor, or human operator—capable of executing one or more processing steps. Resource holons advertise their capabilities and bid for tasks offered by order holons.

Order holon: Represents a customer order or production lot. It is responsible for selecting and sequencing valid processing steps across available resource holons, tracking production progress, and managing exceptions such as machine breakdowns.

Staff holon: An optional advisory holon that provides expert knowledge—such as optimisation heuristics, scheduling algorithms, or historical performance data—to the other holons without assuming direct control authority.

A key design principle of PROSA is the separation of control structure from control algorithm. The same holonic architecture can accommodate both centralised scheduling (by giving the staff holon strong authority) and fully decentralised negotiation, allowing engineers to tune the degree of autonomy to the demands of a particular application.

PROSA was later generalised into the ARTI (Activity–Resource–Type–Instance) reference architecture, which extends the holonic approach beyond manufacturing to logistics, healthcare, and other domains.^[8]

Industrial implementations

Phase 1 of the full HMS Project (concluded February 2000) produced industrial demonstrators at partner facilities:

DaimlerChrysler (Germany): engine assembly cell, reporting a 61% increase in throughput and 37% improvement in robustness.^[9]
Blastman Robotics / VTT (Finland): robotic shot-blasting system.
Toshiba, Hitachi, Fanuc, and Yaskawa (Japan): electric motor assembly.
Alcatel / ATOS (Belgium and Italy): electronic assembly.
GM Holden (Australia): engine machining cell.

Later demonstrators applied holonic principles to die-casting, tyre manufacturing, ceramic tile production, automated guided vehicle (AGV) fleets, and packaging cells.^[5]

Rockwell Automation maintained a long-term research programme in holonic and multi-agent control and published a compendium of its implementations.^[10]

Biologically inspired extensions

Researchers at KU Leuven extended the PROSA architecture with mechanisms drawn from biological systems, particularly swarm intelligence. Pheromone-based shop floor control, inspired by ant colony communication, was used to route orders dynamically through flexible flow shops without centralised scheduling.^[11]

The ADACOR (ADAptive holonic COntrol aRchitecture) system, developed by Paulo Leitão and Francisco Restivo at the Polytechnic Institute of Bragança, introduced dynamic balancing between holarchic and heterarchic control in response to real-time disturbances.^[12]

Relation to multi-agent systems

Holonic manufacturing systems are closely related to multi-agent systems (MAS). A manufacturing holon's control part is typically implemented as a software agent, often using FIPA-compliant agent communication languages. The main distinction is that holonic systems emphasise the dual (autonomous/cooperative) nature of components and explicitly model the recursive self-similarity of the holarchy, whereas general MAS frameworks do not prescribe any particular organisational structure.

The relationship between holonic manufacturing and MAS is surveyed in Leitão, Marík, and Vrba (2013).^[13]

Industry 4.0 and cyber-physical systems

Interest in holonic manufacturing has renewed in the context of Industry 4.0, cyber-physical systems, and the Industrial Internet of Things. A 2020 survey found that holonic manufacturing architectures partially or fully satisfy ten key Industry 4.0 enablers, including decentralised control, reconfigurability, and interoperability.^[14]

Modern extensions include integration with digital twin technology, Manufacturing as a Service (MaaS) business models, and holonic control of AGV fleets. The ARTI+BASE framework (2023) provides a cyber-physical implementation of the ARTI reference architecture. The SOHOMA (Service-Oriented, Holonic and Multi-Agent Manufacturing) workshop series, continuing through 2024, tracks ongoing developments in the field.^[15]

Conference series

The HoloMAS (Holonic and Multi-Agent Systems for Manufacturing) workshop series, published by Springer in Lecture Notes in Computer Science, ran from 2003 to 2011:

HoloMAS 2003, Prague (editors: Marík, McFarlane, Valckenaers)^[16]
HoloMAS 2005, Copenhagen (editors: Marík, Brennan, Pěchouček)
HoloMAS 2007, Regensburg (editors: Marík, Vyatkin, Colombo)
HoloMAS 2009, Linz (editors: Marík, Strasser, Zoitl)
HoloMAS 2011, Toulouse (editors: Mařík, Vrba, Leitão)

The SOHOMA workshop series succeeded HoloMAS and continues to be published by Springer.

References

^ Koestler, Arthur (1967). The Ghost in the Machine. London: Hutchinson.
^ Koestler, Arthur; Smythies, J. R., eds. (1969). Beyond Reductionism: New Perspectives in the Life Sciences. London: Hutchinson.
^ Jump up to: ^a ^b Valckenaers, P.; Van Brussel, H.; Bonneville, F.; Bongaerts, L.; Wyns, J. (1994). "IMS Test Case 5: Holonic Manufacturing Systems". IFAC Proceedings Volumes. 27 (4): 31–36. doi:10.1016/S1474-6670(17)45996-9.
^ Christensen, J. H. (1994). "Holonic Manufacturing Systems: Initial Architecture and Standards Directions". First European Conference on Holonic Manufacturing Systems. Hannover, Germany.
^ Jump up to: ^a ^b Babiceanu, Radu F.; Chen, F. F. (2006). "Development and applications of holonic manufacturing systems: A survey". Journal of Intelligent Manufacturing. 17 (1): 111–131. doi:10.1007/s10845-005-5516-y.
^ Jump up to: ^a ^b Van Brussel, H.; Wyns, J.; Valckenaers, P.; Bongaerts, L.; Peeters, P. (1998). "Reference architecture for holonic manufacturing systems: PROSA". Computers in Industry. 37 (3): 255–274. doi:10.1016/S0166-3615(98)00102-X.
^ Smith, R. G. (1980). "The Contract Net Protocol: High-Level Communication and Control in a Distributed Problem Solver". IEEE Transactions on Computers. C-29 (12): 1104–1113. doi:10.1109/TC.1980.1675516.
^ Valckenaers, Paul (2019). "ARTI Reference Architecture – PROSA Revisited". Service Orientation in Holonic and Multi-Agent Manufacturing. Studies in Computational Intelligence. Vol. 803. Springer. pp. 1–19. doi:10.1007/978-3-030-03003-2_1.
^ Bussmann, S.; Sieverding, J. (2001). "Holonic control of an engine assembly plant: an industrial evaluation". Proceedings of the 2001 IEEE International Conference on Systems, Man and Cybernetics. Vol. 1. pp. 169–174. doi:10.1109/ICSMC.2001.969807.
^ Vrba, Pavel; Tichy, Pavel; Marik, Vladimir; Hall, Kenneth H.; Staron, Raymond J.; Maturana, Francisco P.; Kadera, Petr (2011). "Rockwell Automation's holonic and multiagent control systems compendium". IEEE Transactions on Systems, Man, and Cybernetics, Part C. 41 (1): 14–30. doi:10.1109/TSMCC.2010.2055852.
^ Peeters, P.; Van Brussel, H.; Valckenaers, P.; Wyns, J.; Bongaerts, L.; Kollingbaum, M.; Heikkilä, T. (2001). "Pheromone based emergent shop floor control system for flexible flow shops". Artificial Intelligence in Engineering. 15 (4): 343–352. doi:10.1016/s0954-1810(01)00026-7.
^ Leitão, Paulo; Restivo, Francisco (2006). "ADACOR: A holonic architecture for agile and adaptive manufacturing control". Computers in Industry. 57 (2): 121–130. doi:10.1016/j.compind.2005.05.005.
^ Leitão, Paulo; Marík, Vladimír; Vrba, Pavel (2013). "Past, present, and future of industrial agent applications". IEEE Transactions on Industrial Informatics. 9 (4): 2360–2372. doi:10.1109/TII.2012.2222034.
^ Derigent, William; Cardin, Olivier; Trentesaux, Damien (2021). "Industry 4.0: contributions of holonic manufacturing control architectures and future challenges". Journal of Intelligent Manufacturing. 32 (7): 1797–1818. doi:10.1007/s10845-020-01532-x.
^ Service Oriented, Holonic and Multi-Agent Manufacturing Systems for Industry of the Future. Studies in Computational Intelligence. Springer. 2023.
^ Mařík, V.; McFarlane, D.; Valckenaers, P., eds. (2003). Holonic and Multi-Agent Systems for Manufacturing. Lecture Notes in Computer Science. Vol. 2744. Springer. doi:10.1007/b11833. ISBN 978-3-540-40751-5.