When mentioning autonomous systems today, cars is the immediate association for most people. A lot is today researched and written about that subject. And many authors propose to adapt the concept of autonomy to other technologies, particularly to industrial systems.
To understand what this could mean, let’s explore a little thought experiment. You surely remember the fairy tale of the sleeping beauty. The thirteenth fairy who was not invited to celebrate the birth of the princess cursed the court and the newborn to die. One of the invited fairies then weakened the spell to be a 100 year sleep, after which the princess was to be awoken by a prince.
Now let’s assume that the cursing fairy lives in today’s globalized world and cursed the whole world, i.e. all people on earth. Maybe her powers wouldn’t be sufficient to put that massive number of people to sleep for 100 years, but maybe just one year. But to salvage humanity, one person must stay awake: the prince to awaken the princess and break the spell.
So from one minute to the other, that poor prince is the only person on earth who is awake and active. For the sake of the thought experiment, let’s imagine what his life is like. For some time he is able to eat from the storage from the palace’s kitchen. He can still drive his car as far as the current tank filling takes him. But for many things he wants to do, he needs power, e.g. to cook, or to operate the gas pump. How long would he have power if nobody is operating the power system? Ever since that thought experiment came to my mind, I asked the opinion of power system experts. Nobody was able to tell me, and nobody would even be able to tell how the power system would fail. Building on the fact that the power grid is highly automated and there is power, highly automated factories could still continue to produce for some time without human interaction. But as long as the supply chain is not autonomous as well, supply runs out quickly, and as soon as something fails, even the most advanced factories will shut down.
To survive the year as the only active person on earth would be very tough for the prince.
What can we take from that thought experiment? Autonomous industries are even further away than autonomous cars. Few production sites are automated today to the level where no human reaction is required. And even if they are, handling exceptions is never possible without human intervention. If we interpret ‘autonomous’ as not only independent of human interaction, but independent in a broader sense, we see dependencies all around the plant: it is dependent of energy supply, material supply, maybe water, and it also requires the downstream value chain in place to generate revenues.
But since evil fairies and dormant princesses are rare, some human interaction is acceptable even in a highly automated production environment. The key to success is to figure out where to best use human capabilities, and where the cost of covering all eventualities becomes too high. Furthermore, it is interesting to investigate in which form to deploy the human capabilities, whether it is manual labor and physical presence, or whether automatically operating factories can be supervised and controlled remotely. Humans would then support and supervise the autonomous system, being on the loop rather than in the loop.
This removes people from dangerous, dull, and dirty environments and allows them to work more efficiently. And even if evil fairies are rare, there may be situations where working remotely becomes a necessity from one day to the other. We have just gone through one in the last few months. What was possible for office work, to work from home, would then become possible also for industrial workers.
Autonomous industries may be a vision that is far out, but remotely supervised production may be required again sooner than we think.