Industrial Automation in 2022: From Mechanization to Automation
There was a time, not so long ago, when many of the products purchased by businesses and consumers were manufactured in large factories employing hundreds or even thousands of workers. Such factories were more productive and profitable than earlier manufacturing methods because of mechanization—the enhancement (not replacement) of human labor through powered machines and tools that increased each worker’s productivity.
Mechanization was the primary contribution of Henry Ford’s innovations in assembly-line technology—production-line innovations that enabled Ford to build tens of thousands of automobiles every year, and sell them at prices his factory workers could afford.
Although mechanization increased productivity, humans were, by and large, still in the loop. And humans, sadly, possess a number of counterproductive characteristics:
With automation, human labor is replaced by machines. So, instead of a human madly tightening bolts (like Charlie Chaplin’s character in the 1936 film Modern Times) with a manual wrench or a power tool, a machine automatically tightens the bolts instead, with no human intervention.
Types of Automation Tools
The reduced cost increased computational power of modern computer hardware has enabled the development of a wide range of automation tools, such as:
Supervisory control and data acquisition (SCADA): This tool falls in the supervising and production control category, combining data from lower-level devices and displaying it on easy-to-read dashboards. This tool summarizes the health of one or more manufacturing processes and enables real-time monitoring and control.
Human-machine interfaces (HMIs): Gone are the are the dials, knobs, sight glasses, and gauges of yesteryear; industrial automation devices can now be controlled on PCs, laptops, tablets, and even mobile phones and augmented-reality (AR) headsets.
Artificial neural networks (ANNs): This is a class of artificial intelligence (AI) that is used in machine learning systems. These technologies “learn” by examining thousands of data objects, such as digital images, that are annotated in some way, and forming a mathematical model and adjusting it over time so that the system can label or classify a given data object in a way that matches its annotations.
In an industrial context, a computer vision system can learn to distinguish manufactured items that meet quality criteria from those that do not. Such an automated system could examine every single item, not just a sampling as is typically done by human quality inspectors. By incorporating sensor data, an AI-powered QA system could identify faulty items much more accurately than could a human inspector.
Distributed control systems (DCSs): In this system, process controllers are distributed as closely as possible to the production equipment, rather than centralizing control in one high-level device. This approach increases reliability while still enabling supervisory visibility through HMIs.
Robotics: Industrial robots are becoming increasingly capable and adaptable. Fixed factory-floor robots are easier than ever to program and can be adapted to different products and product lines. Although they have not yet reached the level of understanding natural-language instructions (a la R2-D2), their user interfaces are more intuitive and require less skill to operate than the industrial robots of old.
