What engineers need to know about AI in automationFollow article
Industry 4.0 is constantly evolving, and with artificial intelligence moving closer to the edge, it is set to change even more.
The term Industry 4.0, describing the fourth industrial revolution, was first used in 2015 but was rapidly accepted across the industry. Then, it didn’t really include a specific reference to artificial intelligence (AI), but it is generally felt that I 4.0 is a flexible term that will apply to the ongoing process of digital transformation in the industrial automation sector.
Industrial automation is a fast-paced vertical sector, where cost is always a critical factor. As automation is all about minimising cost, the technology used in industrial automation is constantly evolving. Small gains can equate to large savings, but as automation becomes more complex the supply chain is developing more layers. Specialist providers are developing and adapting their solutions to address these demands. This article provides a glimpse into how that world is developing and offers some insights into how AI is impacting automation and the engineering community.
A short history of automation
We could say that automation started with the first industrial robot in the 1960s. The argument could also be made that automation started on the production line, which dates back much further, to the days of Henry Ford.
Robotics revolutionised production lines, giving rise to the term Robotic Process Automation. This essentially applies to any repetitive process that was simple enough to be carried out by a machine (or software), with little or no human operator intervention.
In a production environment, automation requires a constant flow of raw materials. A robot programmed to process those raw materials may be able to do that task endlessly, as long as the raw materials are made available, and the processed materials are removed. Machine tending describes this process of raw/processed material handling. When robots were introduced, they were static and intended to carry out relatively few tasks. They needed tending by skilled operators. As they have developed, the need for skilled operators has diminished. Today, some of the machine tending tasks are likely to be carried out by other robots.
This trend is accelerating with the introduction of AI. This has given rise to the concept of ‘lights out manufacturing’, where there is very little or even no human presence, hence there is no need for lighting.
Five key developments
There are many areas where AI and automation are beating new paths. This could continue to the point where we can enjoy on-demand manufacturing carried out by lights-out factories operating on energy generated from entirely renewable sources, using fully recycled and recyclable materials.
In order to get to that point, several key technologies are required. All of these areas are going through their own digital transformation, due to the introduction of artificial intelligence.
Intelligent Process Automation: This describes how robotic process automation (see above) is transforming due to the introduction of AI. Effectively, robots are no longer constrained to performing the same task in the same way over and over again. They may still perform the same task, but they will be empowered to vary and improve the way they perform the task based on external conditions or other influences. This may be as simple as being able to detect a slight defect in a raw material using machine vision (see below).
Cobots: Collaborative robots, or Cobots, are robots that have been designed specifically to work alongside a human operator. This is significantly different from robotic process automation, or even intelligent process automation because it requires the robot to be more aware of its surroundings. More importantly, a Cobot needs to be designed to be safe in the presence of humans. Many industrial robots operate within caged areas, to protect humans from wandering into the robot’s operating radius. With a Cobot, there may be no physical barrier between it and the human. This will use technologies such as AI to detect potential hazards and react to them safely.
Machine Vision: while not a new technology, machine vision is getting an upgrade thanks to AI. Adding cameras to production lines is commonplace, but putting AI inside the cameras, or using AI to analyse what the cameras are seeing, takes machine vision to a new level.
Automated Mobile Robots: While Cobots are intended to operate alongside people, they may still be static. Automated mobile robots, or AMRs, are designed to move and operate within a much larger area. As well as controlling its functionality, AI will be used in AMRs to detect its location, its surroundings and, of course, detect potential safety hazards.
Edge processing: This is a technology that will revolutionise more than just industrial automation. As AI and machine learning find their way into more devices, those devices will be located nearer to the edge of the network. Smart nodes in the IoT will be one form of edge processing, but so too could smart machinery.
The future of AI in automation
Artificial intelligence comes in many forms. It can bring expertise and consistency to automated tasks, making them more efficient. It can also allow systems to learn, which can lead to a greater capacity to handle variability.
The repetitive application of experience without variation, combined with the capacity to continuously learn and improve, is unique to artificially intelligent systems. This is what AI can bring to industrial automation in the future.
AI is expected to give all industries an uplift in terms of growth over the coming decade. Analysts seem to be united in their belief that AI will provide somewhere in the region of 20% additional growth in that period.
We can expect AI to become more integrated into industrial automation as the technology and the techniques develop together. The technology is based on both the hardware and the software, which will be widely applicable across all verticals. The techniques for implementing that technology will be tuned to the application.
For this reason, we can expect AI to penetrate specific automation tasks first. This is already happening, with the use of machine learning in vision systems used for product inspection, for example.
We should not underestimate the depth to which AI will penetrate automation. In the near future, AI could be running the majority of industrial processes, operating at multiple levels and creating a new paradigm in manufacturing.