Artificial intelligence (AI) has had a bad press. In the past two decades, this technology has gradually disrupted the world of employment as jobs have been decimated by its relentless march.
From its depiction in Hollywood movies to media scare stories to widespread public perception, AI has a reputation that belies its actual influence on improving and streamlining human lives.
For every job lost to AI, the technology has slashed costs in manufacturing where humans ultimately benefit from lower costs. If it wasn’t for AI, then humans would be paying much more for the goods that turn up on their doorsteps. It’s a weird kind of quid pro quo.
This rise of the machines in all areas of life has been matched by the machine revolution in manufacturing. What were once nascent (and relatively primitive) robots on assembly lines have become swingeing means of mass production.
While humans are still coming to terms with how their jobs are going to be able to cope with machines, let alone artificial intelligence (AI), those who do understand AI are using the less baleful elements of this technology to revolutionise manufacturing… in a positive way.
AI’s use of automation in the manufacturing industry takes many forms and for every single manufacturing company, there are several pivotal points where automation comes to bear.
These range from distribution of warehouse products, a supply chain that involves a number of multiple suppliers, to placing orders themselves. At any point waste can be eliminated or, at the very least, improved by automation.
Resource and material distribution are two areas where automation powered by machine learning can automatically streamline and reorder supplies. This works for last-minute manufacturing or juggling resources from adjacent facilities when one facility has a surplus; this subsequently prevents local troughs of supply and demand.
The supply chain is more complicated, but this is where automation plays a strong hand. Multi-vendor workflows mean manufacturers rely on different partners for raw material sourcing, outside testing, clearing customs, filing paperwork, exchanging currencies and much more.
Ongoing Brexit confusion means that automation may prove to be even more essential to UK (and European) companies when it comes to decisions that can be made without human intervention such as route planning and other bureaucratic transportation challenges.
Finally, a more AI-driven approach to the management of suppliers mean manufacturing companies can ‘insure’ the future. No human knows which way the wind will blow, but automated AI solutions mean that outlier weather conditions can be predicted in advance, although very few foresaw the ultimate outlier - Brexit.
Then there is the increasingly important area where AI literally comes to its senses and that is the ability of machines to ‘see’ for themselves. Image recognition or object detection means that vision-enabled robots can read products from an image and can even see the presence of humans.
This ability can be split into two parts. Computer vision gives computers the ability to ‘see’ living objects such as humans while object detection helps find and locate specific objects within an image.
For detecting generic objects there are open-source and pre-trained models available, but if an algorithm is needed to identify specific objects, it is possible to create and ‘teach’ a proprietary object detection algorithm.
Humans, of course, can see for themselves and humans constantly adapt their vision to deal with parts of the manufacturing cycle that rely on this sensitivity to unexpected changes.
Using AI, however, means many such challenges can be handled. AI may be non-human but because of its speed in automatically distinguishing good parts from faulty parts on a dynamic assembly line, means it can be more effective than a slow-reacting human.
While not every human worker is perfect, neither is AI, but AI-powered object tracking means objects can be classified in ways selected by the manufacturer and where the statistics of this number of objects is pre-ordained.
Consequently, it reduces discrepancies in categorisation and creates a more flexible and proactive assembly line. For tedious and manual tasks such as sorting and discarding products, object detection transforms this provides a higher level of accuracy and a concomitant efficient and automated process.
Finally, image recognition helps to improve inventory management because of the difficulty in tracking products in real-time, another constantly changing process where units are every day added, subtracted or moved to different parts of the assembly line.
There can be no diminution of how important inventory management is and how it can badly affect a company’s efficiency. However, with AI image recognition, automatic object counting and localisation make inventory infinitely more accurate.
Not only is (understandable) human error removed from the process, automated businesses can order the right quantity of products at the best possible price, ensuring less waste and more productivity.
Today there are fully automated assembly lines for complex products such as cars. Each robotic arm movement is scripted by AI meaning materials and components are commensurately defined. As AI evolves, machines will use AI-powered object detection to continually improve.
As the early 21st Century evolves, the shelf life of products will focus on further AI automation and image recognition evolution. This is a manufacturing revolution that has already happened, but one that is still to play out. At every level, the consequences will be seismic and the earth will move for everybody.
To enable comments sign up for a Disqus account and enter your Disqus shortname in the Articulate node settings.