The Autonomous Factories of the Future
How Decentralisation, AI, Blockchain and Cryptocurrency unlock the true potential of the Industry 4.0 Revolution
An OT Industrial Control Engineer for Thales shares their vision for the future of autonomous factories…
In 2018, Tesla's fully autonomous Model 3 factory was severely underperforming and fraught with difficulties. The factory, built in 2016, was designed to manufacture 5000 cars per week, but was struggling to manage 2000. Tesla was under fire for the manufacturing delays and faults in their vehicles that resulted in a large number of recalls. In the end, certain parts of the automation processes were removed and people were brought back into the factory. The fully autonomous factory was arguably a mistake at the time but what the world learnt from Tesla's Model 3 factory was that a greater level of factory automation was achievable and since then, technology has moved on even further.
Many industries are experimenting with increased levels of automation, a trend that has been accelerated by the COVID-19 pandemic. Full autonomy will provide enhanced efficiency, safety and quality. It frees up the employees to do creative work such as engineering, product design and building relationships with suppliers and customers.
To visualise a potential application for full autonomy, imagine a pharmaceutical factory producing personalised medicine, where every packet of tablets is specifically formulated to cater to an individual patient's needs. The logistics of this are incredibly complex. The patient's doctor would need to specify the medication in detail, the pharmacist would need to order that medication on a regular basis, the factory would need to produce that medication and send it to the pharmacist on time. Throughout this chain, it would be of vital importance that patient details are not confused and their privacy is preserved. From an operational perspective, the factory’s machines would need to be capable of manufacturing the different prescriptions for each patient. In addition, the pharmaceutical factory would be highly regulated, and need to provide full traceability of the ingredients and manufacturing techniques for each batch of medicine.
How could the challenge be addressed?
If the logistics firms carrying out the distribution of medicines had a single, centralised server conducting AI calculations, then this would represent a single point of failure. If the server went down, there would be catastrophic consequences. Hosting the AI in the cloud reduces the likelihood of this event, but as the Fastly outage showed, even the cloud isn't infallible.
One potential option is to make every vehicle in the logistics organisation an AI node that was capable of sharing data with every other vehicle. By this means, the vans would create a distributed AI that had no single point of failure and this thinking could also be applied to the other elements within the supply chain, such as the pharmacies and factory machinery.
Such a method, hypothetically, would create a complex web of communications, where all the AI nodes are working together to drive maximal efficiency within the entire supply chain. From a traceability point of view, however, unpicking that web would be a nightmare. What if, therefore, all of the communication was registered on a blockchain? If all of the data exchange was recorded in an immutable ledger, then it would be possible to trace the process of the prescription manufacture right from the source of the ingredients, through to the pseudonymous reference of the driver who delivered it. This data could, in theory, be accessible by a QR code printed on the packet that linked to a webpage collating all of this transaction data.
If the machines were also capable of handling monetary transactions, then the purchasing process could be autonomous too. In order to achieve this, the relevant purchasing department would need to develop a white list of suppliers that meet the required standards and regulations. All transactions could then be handled in cryptocurrency in order to take advantage of the blockchain ledger, reduced transaction fees and improved security.
This is how I imagine the fully autonomous factory, and its associated supply chain supply chain, working. These principles could be applied in any manufacturing industry, although trying to build this right now could be a mistake equivalent to Tesla’s factory in 2016. I do not, however, think that this future is too far away.