Programmable Logic Controllers (PLCs) have been central to automation systems since the 1980s. These devices enabled the sequencing of inputs and outputs and supported proprietary bus systems that accelerated the adoption of programmable automation. Early machine vision systems relied on simple input triggers and output pass/fail signals, with measurements sometimes delivered via serial or even analog outputs.
As networking technology gained traction in the IT world around the turn of the millennium, PLC vendors responded by introducing real-time industrial Ethernet protocols like Ethernet/IP, ProfiNet, and EtherCAT. These innovations allowed automation networks to function in real time, enabling greater distances between devices and significantly increasing the volume of data that could be exchanged.
Soon after, in 2006, a major shift occurred in the machine vision sector. With the reduction in power consumption of network chips, compact cameras could utilize Ethernet without overheating. This led to the launch of GigE Vision by the AIA (now the A3), a standard for camera control and video streaming via Ethernet. Just like PLC manufacturers who leveraged mass-market technology to reduce costs and increase functionality, machine vision systems benefited from faster speeds, smaller sizes, longer cable lengths, and lower costs.
A key feature of both PLC and machine vision networking was the creation of isolated operational networks—separate from company-wide IT networks to maintain performance regardless of network load. However, early users often mistakenly connected high-bandwidth vision systems to corporate networks, occasionally bringing the entire system to a halt due to high data demands.
This isolation of Operational Technology (OT) networks from IT networks became a critical design principle. However, friction often arose when IT departments attempted to manage OT infrastructure. For instance, deploying Windows updates to a vision system PC while a production line was running could inadvertently halt manufacturing. In the early days, these systems were often completely segregated from corporate networks, reducing risks from viruses or external security threats. This made it easier to negotiate limited IT oversight.
The landscape shifted again as equipment vendors began offering remote access capabilities. This allowed engineers to diagnose issues without site visits, improving uptime and reducing maintenance costs. However, this new connectivity introduced valid security concerns, as unauthorized access could potentially lead to data theft, sabotage, or system hacking.
As we entered the 2010s, the vision of the Smart Factory and Industry 4.0 began to take hold. Automation and vision systems were now more connected and configurable. Some large manufacturers ramped up product personalization, with production lines dynamically reconfiguring so each item was unique. This shift required seamless data flow from ERP (Enterprise Resource Planning) to MRP (Manufacturing Resource Planning) systems, down to the shop floor.
At a VDMA Machine Vision Summit in Mallorca, two powerful examples illustrated this transformation:
- BMW implemented a factory-wide deep learning vision system to verify each car’s configuration during assembly—checking for correct wheels, badges, colors, and interior features. The system relayed build status to the MRP, ensuring every vehicle matched its order.
- Nike showcased its “Nike By You” service, where customers could design their own trainers online. Each unique configuration was transmitted to the factory for production at a quantity of one.
While these are high-profile examples, Industry 4.0 also includes more modest implementations. These could be as simple as linking quality reports to individual serial numbers or recording production batch statistics—like reject counts and reworks—for long-term material tracking and process improvement.
To make this level of integration possible, every element in the factory needs to communicate. This led to the development of OPC UA (Open Platform Communication Unified Architecture), a standardized, secure, and scalable communication protocol. OPC UA spans everything from cloud-based ERP/MRP systems to low-level real-time controllers, offering a modern, cybersecurity-ready alternative to traditional fieldbus protocols.
Various industries have adopted OPC UA through tailored “companion specifications.” In 2018, the VDMA Machine Vision group launched the OPC UA Vision specification, which continues to evolve. It defines standardized ways to control and interact with any vision system, helping unify disparate technologies. A helpful YouTube presentation about this standard can be found here.
To further improve factory visibility, the Universal Machine Technology Interface (UMATI) was introduced. UMATI facilitates data exchange between machines and their integration in to customer specific IT ecosystems, based on OPC-UA. It provides insight into machine status, performance metrics, recipes, health checks, and predictive maintenance—all in a unified, interoperable format. More details are available at umati.org.
Despite the potential, Industry 4.0 adoption has been slower than anticipated. Key barriers include cybersecurity concerns, infrastructure investment and the need for skilled personnel. In addition, as many factory production lines have a life in excess of 10 years the challenge to integrate new and old technology increases the integration complexity further dampening the viability of adoption.
The concern over cyberattacks has made IT departments more involved in OT systems. Fortunately, there’s growing recognition that OT requires a distinct approach.
To ensure these connected environments remain secure, the globally recognized IEC 62443 standard offers a comprehensive framework. It covers not only the technical requirements for both IT and OT but also emphasizes secure design, implementation procedures, operational policies, and ongoing training for all personnel involved.
A further step to help secure networks is the new EU Cyber Resilience Act which was adopted in October 2024 introduces cybersecurity requirements for products with digital elements starting in December 2027. While this is EU only its expected similar legislation will be introduced globally. This may push further adoption of protocols like OPC UA, which offer embedded security, unlike many legacy fieldbus systems still in widespread use and will force a shift to the new technology.
With the drive to reduce operational costs, maintaining automation and vision systems efficiently has become increasingly critical, especially when integrated directly into production processes. As the use of artificial intelligence becomes more common in the workplace, more professionals are realizing the potential value of big data collected from production machines which will drive the demand to capture this data and hence increase the value of connected machines.
The move toward Industry 4.0 will no longer just the domain of large manufacturers. Thanks to maturing interoperability standards, emerging security regulations, and proven implementation practices, mid-sized companies will have a clearer and safer path to adoption.
The path to the true smart factory is there and while the adoption has been slower than first expected the building block are coming together to make useful implementations easier and safer to deploy.
