The manufacturing industry is undergoing a profound transformation, evolving from rigid automation toward autonomous, adaptive, and resilient production systems. This shift is driven by the volatility of global markets, the need for sustainability, and the growing shortage of skilled labour.[1]This gives rise to the so-called self-adapting factory: an environment where industrial assets not only perform tasks, but also perceive their surroundings and respond dynamically to them[2],[3].
Factories must now not only produce, but also adapt: to supply chain disruptions, product customization, and the variability of raw materials (especially relevant in bio-based industries). to energy crises or unexpected global events. In this context, digital transformation cannot be limited to incorporating isolated technologies, but must generate real capabilities for autonomy and resilience, keeping the human being at the centre, as promoted by the Industry 5.0.
From classic control to intelligent autonomy
Traditionally, industrial automation has relied on fixed rules programmed into industrial controllers, effective in stable environments but vulnerable to variability. In contrast, intelligent automation introduces artificial intelligence and machine learning algorithms capable of learning from process behaviour and adjusting parameters in real time [4],[5].
In this context, self-X capabilities emerged, enabling industrial systems to autonomously monitor, configure, recover, and optimize themselves. These capabilities were the core of the European s-X-AIPI project coordinated by CARTIF, whose objective was to provide the industry with more robust, reusable and reliable AI solution processes.
Digital twins: from analysis to action
One of the key enablers of this autonomy is the digital twin[6]. A digital twin is understood as a dynamic virtual representation of a physical asset or process synchronized with real-world data. Unlike a static simulation, a digital twin accompanies the system throughout its entire lifecycle and allows for testing strategies without risk to production.[7].
The combination of digital twins with artificial intelligence transforms these models into operational tools: they allow for anticipating deviations, optimizing parameters, and supporting real-time decision-making. As a result, optimizing processes becomes one of the most transformative impacts of industrial digitalization, enabling significant improvements in sustainability efficiency and energy consumption, even in contexts of high variability.
Horizon 2026: Agent AI and Interoperability
If industrial AI has enabled analysis and optimization, the next step is to move towards new types of AI with greater autonomy, capable of pursuing objectives and coordinating complex actions without constant supervision. This approach is known as agentic AI [8] and it will be key from 2026 onwards.
For this paradigm to be viable in real-world industrial environments, interoperability is critical. In this regard, the Asset Administration Shell (AAS) is positioned as the standard that allows assets from different manufacturers to share information using a common language. This same technological foundation will be essential for the deployment of the Digital Product Passport (DPP), which will be mandatory in sectors such as batteries from 2027 and will require traceability, transparency and a circular economy.
In this context, CARTIF is participating in the bi0SpaCE project, focused on developing digitalization solutions and data spaces for bio-based industries, facilitating raw material tracking, process monitoring, and adaptation to future regulatory requirements. Adaptation is no longer just technical, but also regulatory and strategic.
Data and cybersecurity: the invisible challenge
Despite the progress, the transition to autonomous factories faces structural challenges. Data silos continue to fragment business intelligence, consuming a significant portion of development effort on data preparation tasks. Added to this are emerging threats such as data poisoning which can severely degrade the performance of predictive models. Given this scenario, adopting Zero Trust architectures is essential to guarantee the integrity, traceability and reliability of the data and models on which industrial autonomy is based.
Collaborative robots: adaptation is also physical
An adaptive factory not only learns and predicts, but also acts. Collaborative robotics is thus becoming another pillar of Industry 5.0, enabling safe interaction between people and robots in shared environments.
CARTIF contributes to this progress through the European project ARISE, which develops open middleware and reusable tools to facilitate the integration of collaborative robotics in industry. The goal is to reduce technological barriers and accelerate the adoption of flexible, interoperable, and human-centered robotic solutions.
The human factor in Industry 5.0
Autonomy does not eliminate human beings, but rather transforms their role. As industrial systems become more intelligent, people evolve into more specific functions[9] like strategic supervision, decision-making and management of complex scenarios. This gives rise to new professional profiles linked to automation, AI ethics, and the orchestration of hybrid human-machine systems.
In this context, industrial autonomy should not be interpreted as a replacement for the human factor, but rather as a natural evolution of the worker’s role within the factory.
Conclusion
The factory of the future, which is drawing ever closer, will not only be digital, but also autonomous without losing the human element. Industry 4.0 enabled connectivity and digitization. Industry 5.0 now demands adaptation, optimization, and humanization.
Digital twins, trusted industrial AI, collaborative robotics, and advanced traceability are all pieces of the same puzzle: building factories capable of responding nimbly to a changing world. At CARTIF, we work to ensure these technologies are not just concepts, but real tools that drive competitiveness, sustainability, and resilience in the European industrial fabric. The self-adapting factory is no longer science fiction; it’s the next step in the industrial revolution.
[1] RPA: a strategic solution to the challenges facing the manufacturing industry https://www.rautomation.es/2025/03/17/automatizacion-robotica-procesos-rpa-solucion-estrategica-desafios-industria-manufacturera
[2] Sustainable Transformation in the Manufacturing Industry: Synergies between Agility, Artificial Intelligence, and Change Resistance Management https://www.reincisol.com/ojs/index.php/reincisol/article/view/918
[3] Industry 4.0, the new driver of industrial innovation: https://www.revistadyo.es/DyO/index.php/dyo/article/view/563
[4] AI and machine learning applied to industrial engineering to improve operations management: https://revistas.ulacit.ac.cr/index.php/rhombus/article/view/343
[5] Analysis of the use of machine learning for predictive control systems at the industrial level: https://polodelconocimiento.com/ojs/index.php/es/article/view/7549
[6] Digital Twin Manufacturing: Applications, Benefits, and Industry Insights https://www.simio.com/digital-twin-manufacturing-applications-benefits-and-industry-insights/
[7] Digital Twin in Manufacturing https://www.autodesk.com/blogs/design-and-manufacturing/digital-twin-in-manufacturing/
[8] 10 Agentic AI Examples and Use Cases https://boomi.com/blog/10-agentic-ai-use-cases/
[9] 4 ways artificial intelligence could transform manufacturing https://www.weforum.org/stories/2023/01/4-ways-artificial-intelligence-manufacturing-davos2023/
More about Industry 4.0…
