Skip to main content

AI & Robotics in the Steel Industry: When Machines Talk to Each Other

© Adobe Stock / InfiniteFlow

Smart industry robot arms for digital factory production technology showing automation manufacturing process of the Industry 4.0 or 4th industrial revolution and AIoT software to control operation (Source: Adobe Stock / InfiniteFlow)

AI is now making broad inroads onto the shop floor throughout the steel industry: The Internet of Things (IoT) is evolving into an Artificial Intelligence of Things (AIoT) – intelligent, networked systems that not only collect data but make decisions autonomously. Increasingly more steel producers worldwide are now taking the next step toward plant autonomy, where machines communicate with each other, learn from one another, and optimize production processes in real time. 

 

The transformation to intelligent autonomous production is no longer a vision of the future – it is already in full swing. In 2025, 20 billion devices will be connected to the IoT worldwide; by 2033, this number is expected to more than double to over 40 billion. The market for connected devices reached a volume of 1,020 billion euros in 2025. 

Even more impressive is the development in the AIoT sector: the global market here is growing from 171 billion US dollars (2024) to a projected 897 billion US dollars by 2030 – an average annual growth rate of over 31%. 

The picture is similar in robotics: in 2024, 542,000 new industrial robots were installed worldwide – more than twice as many as ten years ago. China leads with a share of 54% of all units and reached an operational stock of over 2 million robots in 2024. Germany remains Europe's largest robot market. During the same period, in 2024, there were 27,000 new installations here. 

And another survey result is clear: 78% of companies worldwide plan to increase their AI investments in 2026. The question is no longer whether, but how quickly companies implement these technologies. 

 

AIoT: Decentralized Intelligence on the Shop Floor 

At the top of the AI wish list is the networking of various assets on the shop floor, meaning the interconnection of machines, robots, cobots, drives, actuators, sensors, and all control and regulation technology. The classic Internet of Things (IoT) is increasingly being expanded with AI functions and is transforming into the Artificial Intelligence of Things (AIoT) – a system that goes far beyond mere data collection. 

 

How Machines Talk to Each Other 

The true revolution of AIoT lies in decentralized intelligence: instead of processing data centrally in a cloud, analysis increasingly takes place where it originates – directly at the machine or at the network edge. This Edge AI enables production facilities to react to changes in the process within milliseconds. 

In modern steel mills, rolling mills, furnaces, cranes, and quality management sensors are in permanent communication with each other. An example: if an AI-supported camera system at a rolling mill detects a surface irregularity, it can not only sound an alarm – it automatically informs upstream process steps about the probable cause and adjusts parameters in real time. Downstream systems simultaneously receive information to treat potentially affected material specially. 

© Adobe Stock / everythingpossible

Analyst working with Business Analytics and Data Management System on computer to make report with KPI and metrics connected to database. Corporate strategy for finance, operations, sales, marketing (Source: Adobe Stock / everythingpossible)

  1. This networked intelligence is based on standardized communication protocols such as OPC UA (Open Platform Communications Unified Architecture), which enable cross-manufacturer data exchange. Every machine becomes a "Smart Asset" that not only knows its own status but understands the context of the entire production process. 

     

    Energy Efficiency Through Intelligent Networking 

    The networking of machines with AI brings measurable benefits in energy efficiency and resource consumption. AIoT systems analyze the energy consumption of each individual plant in real time and coordinate production steps to avoid peak loads. 

    ArcelorMittal, for example, was able to reduce scrap by 20% through AI optimization at the Hamburg wire rolling mill, which directly led to corresponding energy and CO₂ savings. Intelligent systems recognize inefficient operating conditions and automatically suggest optimizations – or even implement them autonomously. 

    At South Korean steel producer POSCO, the Smart Factory platform PosFrame, specially designed for continuous manufacturing processes, led to cost savings of 250 billion Korean won (around 150,000 euros) over four years through 321 so-called Smart Tasks. The AI-controlled blast furnace control system, based on deep learning and over 30 years of operational experience, increased daily pig iron production by 240 tons. 

     

    Intelligent Inspection and Maintenance 

    While classic cranes and welding robots have been remotely controlled and deployed with high degrees of automation in steel mills for decades, AIoT brings a completely new dimension of automation: autonomous systems that move independently through complex industrial environments while continuously learning. 

    Outokumpu was the first steel company worldwide to deploy the four-legged ANYmal robot from ANYbotics for safety management. Three of these robots are now deployed at sites in Germany, Sweden, and Finland. They autonomously conduct inspections in hazardous areas – such as in acidic environments or near high-temperature facilities. 

    The measurable successes are impressive: employee exposure to hazardous substances decreased by more than 80%. Potential repairs in maintenance areas were reduced by 20%. The robots detect anomalies such as overheated bearings or motors early and thus prevent fires and unplanned downtime. 

© Adobe Stock / Just Jus

Boston Dynamics Spot the Dog Robot (Source: Adobe Stock / Just Jus)

POSCO expanded this concept with the Boston Dynamics Spot robot in its Smart Factories. The robot tolerates extreme temperatures, dust, and chemical stress and autonomously monitors highly complex production environments. Its built-in sensors – thermal imaging cameras, gas sensors, high-resolution cameras, and acoustic detectors – continuously capture plant conditions. 

 

Predictive Maintenance: From Reactive to Proactive Action 

The true potential of networking is particularly evident in predictive maintenance. AIoT systems permanently collect operating parameters such as temperature, vibration, pressure, current consumption, and acoustic signatures. This data is analyzed through machine learning – algorithms such as neural networks, random forests, or gradient boosting detect wear patterns long before failure threatens. 

ArcelorMittal set a new industry standard with its Sentinel platform: in the steel manufacturer's pilot plants, the system predicts motor failures and problems with hydraulic actuators with a success rate of 100%. ThyssenKrupp Steel reduced unplanned plant downtime by 20% through AI-supported predictive maintenance systems. 

The crucial advantage: the systems schedule maintenance work for already planned downtime periods. Instead of having to react in emergencies, maintenance teams can operate optimally prepared and equipped with the right spare parts. 

 

Quality Through Intelligent Networking 

AIoT has also initiated a paradigm shift in quality control. High-resolution cameras and specialized lighting systems capture surfaces during production. Deep learning models, particularly Convolutional Neural Networks (CNNs), analyze these images in real time. 

ThyssenKrupp Steel developed deep learning models for surface inspection of packaging steel that automatically detect defects and categorize their causes. These systems often achieve significantly higher detection rates than human inspectors – and work tirelessly with consistent quality, 24 hours a day. 

POSCO implemented AI systems for coating weight control in continuous hot-dip galvanizing lines for automotive steel. The technology combines AI-based prediction models with optimization algorithms for real-time control. Result: drastically reduced deviations in coating weight, consistent product quality independent of operator qualification, and minimized zinc waste. 

 

Autonomous Process Optimization 

AIoT systems not only optimize individual machines – they also orchestrate entire production chains. So-called reinforcement learning enables systems to develop optimal control strategies through trial and error. In real time, the systems adjust process parameters such as temperatures, rolling speeds, or material compositions. 

ThyssenKrupp Steel implemented the Central Operation Cockpit (COC) from Primetals Technologies, which enables control of an entire plant from a single operator point. Intelligent, AI-supported assistance systems give operators a complete overview of the entire production process – a milestone on the path to the fully autonomous plant. 

 

Autonomous Logistics: The Next Level of Networking 

While classic autonomous vehicles often travel along fixed routes, modern AIoT transport systems use a combination of multiple technologies: lidar sensors, radar, cameras, and GPS together provide a complete picture of the environment. Sensor fusion algorithms combine this data for reliable perception. 

© Adobe Stock / Connect world

Packages are transported in high-tech settings, online shopping, concept of automatic logistics management (Source: Adobe Stock / Connect world)

Based on this technical foundation, Nippon Steel developed autonomous heavy-duty transporters for steel mill operations in collaboration with TIER IV. The vehicles navigate autonomously through extensive plant sites and transport steel coils, slabs, and other heavy materials. What's special: they continuously communicate with a central control unit that coordinates transport orders, avoids collisions, and optimizes material flow. 

The vehicles use machine learning models to recognize and classify various objects – employees, other vehicles, material stacks. They dynamically adapt their routes to changes in the environment and seamlessly integrate into ongoing production processes. 

 

Challenges and Solutions 

The transformation to intelligent production involves challenges. High implementation costs for industrial robotics and AIoT systems represent a barrier particularly for medium-sized companies. The integration of different data sources from legacy and modern systems requires considerable effort. 

Companies address these challenges through: 

  • Gradual implementation with pilot projects 
  • Interdisciplinary teams of data scientists and domain experts 
  • Use of open-source technologies for cost reduction 
  • Intensive employee training and change management 

POSCO was recognized in 2019 by the World Economic Forum as the first South Korean "Lighthouse Factory" and demonstrates with a wealth of best practices how the integration of Industry 4.0 technologies can succeed convincingly and with economic success. The company also established an AIoT ecosystem in which SMEs and start-ups can thrive through active collaboration. 

 

Energy Requirements of AI Systems 

An increasingly relevant aspect is the energy demand of AI applications. Training complex deep learning models requires significant computing power. In an industry that is already energy-intensive and pursuing decarbonization goals, this additional energy demand must be carefully considered. 

Companies meet this challenge through: 

  • Use of energy-efficient hardware such as specialized AI chips 
  • Use of cloud computing with optimized infrastructure 
  • Training smaller, efficient models for Edge AI 
  • Prioritization of AI applications with demonstrable benefits 

The net effect is nevertheless positive in many cases: the energy savings achieved through AIoT in production more than compensate for the consumption for AI operations in the vast majority of cases. 

© Adobe Stock / Deemerwha studio

AI artificial intelligence tool analysis stock market. Investor use smart robot automation technology for analysis graph data for decision trading. Robotic arm pointing at stock chart (Source: Adobe Stock / Deemerwha studio)

Outlook: The Autonomous Factory 

The integration of AI and robotics in the steel industry is in a phase of accelerated growth. The following technology trends will provide further boost in the coming years: 

  • Fully autonomous production systems with self-optimizing processes 
  • Digital twins for simulation and optimization of entire plants 
  • Advanced humanoid robots for complex manual tasks 
  • 5G/6G integration for even faster communication between machines 
  • Quantum computing for highly complex optimization problems 

The economic imperatives – cost pressure, quality requirements, skilled labor shortage, sustainability goals, and global competition – make digitalization a strategic necessity. Companies that invest early in intelligent, networked manufacturing systems secure decisive competitive advantages in productivity, product quality, faster response to market requirements, as well as lower resource and energy consumption and improved workplace safety. 

The future of the steel industry will therefore be shaped by machines that not only work but talk to each other, learn from one another, and continuously improve together. 

 

Learn more about AI & Robotics at the next wire & Tube, which will take place from April 13-17, 2026 in Düsseldorf. The internet portal at www.wire.de provides you with current industry and product information. 

 

Further Information and Sources 

  • International Federation of Robotics (IFR): World Robotics Report 2025, https://ifr.org 
  • atp magazin, Issue 3/2025, pp. 16-17. Vulkan-Verlag GmbH, Essen. 

 

Return to top