We’re witnessing the birth of the Factory of the Future where Artificial Intelligence (AI) enables the Internet of Things (IoT). It’s a fast-paced landscape, changing the way we do things, make things, the way we think and the skills we need.
Who better to speak to than BAE Systems’ Nick Colosimo and Alex Griffiths. As Lead Engineer, FCAS Technology, Nick is well placed to give an overview on these new technologies. Just to underscore their subject matter expert status, Nick is a visiting professor at Cranfield University and has supported both the UK and MOD and the European Commission on AI matters while as Senior Integration Engineer, IoT & Artificial Intelligence, Alex is immersed in a new role coordinating the activity around the IoT and applied AI.
The game changer
“I believe AI is one of the most significant inventions mankind will ever make,” says Nick. For a scientist that’s quite a bold statement. However, Nick is clear, while there are a host of other exciting and disruptive technologies emerging during our technological revolution, AI is historic.
That’s an audacious claim.
“Agreed, and as Carl Sagan said: ‘Extraordinary claims require extraordinary evidence’. The thing is many technological inventions are limited in terms of what they may lead on to. Artificial Intelligence is different. Certain types of AI allow you to create new things, to think new things up.
“Nor is AI limited down a particular narrow technology path. In principle, it can manifest in a whole range of different things; what we produce, how we produce it and crucially it will act as a catalyst for many more inventions.
“The Internet allows us to create more through the sharing of information. AI achieves the same, but it can also become a feature of a product. For example, it can act as a decision aid in a next-generation cockpit to advise pilots. At the same time AI can be applied to how we produce, engineer and manufacture things.
“AI is also able to accelerate discovery. No other technological invention besides the computer can do that. This is the next level.”
The jobs myth
But what about the threat to jobs the AI revolution will bring? Nick agrees there will be an impact but believes it’s not quite all doom and gloom.
He says: “For one thing AI means there are new jobs in areas like data science, data engineering and AI engineering, plus a whole range of new products which involve levels of autonomy. That in itself creates new job roles which never existed before.”
But he says AI will change how we carry out our jobs — often for the better.
“For example, if you're an engineer running millions of computational fluid dynamics cases — the aerodynamic models which allow us to optimise things like wing design — AI may be able to analyse your stated intent and tell you it can get the answer you need by just running 20 test cases rather than 500. It's that sort of insight and intelligence which we will start to see impacting us all in our jobs over the next ten years.”
Nick points out that AI is already helping us in our daily lives. When we watch Netflix, AI is used to make suggestions about what to watch next.
“It's already becoming ubiquitous. But over the next ten years, we'll see a much more direct impact into how we complete our jobs. What I don't envisage for many decades is what’s called artificial general intelligence — that’s AI that’s as smart as anyone reading this article. But rather we will have AI that is better than humans in very narrow applications.”
Transforming the factory
In essence, the IoT is micro-electronics which are able to sense their environment, and then communicate that out. The IoT applies to almost any device. At home, a Nest thermostat is a simple example of the IoT.
In our work it might apply to components in the supply chain each tagged with small devices that can communicate with the Internet. You then know exactly where they are at any one point in time and their health status. Is it a healthy or unhealthy package? Has it been dropped? Has it been moved recently? Gaining this knowledge can be transformative within a manufacturing environment.
Data is the new oil
“AI thrives on data,” says Alex, who is currently involved with installing the IoT network into BAE Systems’ demonstration factory. “There are few things more important in generating and training successful AI than having data for it to learn from. That’s why people say data is the new oil, because it’s such an incredibly valuable asset.”
The IoT within a factory environment will allow us access to a great deal more data and a great deal more varied data.
“It’s often dubbed Industry 4.0,” says Alex. “It’s about moving from a factory being a building full of machines to the factory being one singular machine in which everything is a part, even if it is not physically connected. Every machine and the sensors within them feeds data back in. It could be as simple as a sensor on a component making sure it’s not exposed to high heat, or as complex as a robot system, feeding back all kinds of information about its position, the tools it has, the job it is doing.”
AI in this environment has a huge amount of data it can learn from, which means it can have a significant impact in terms of optimisation.
“One of the more basic implications is of AI for maintenance. It can spot when the machine is going to go wrong before any human operator. Instead of unscheduled down time when a machine breaks you can schedule the servicing. That's one of the things that we are actively pursuing.”
Nick gives another example: “The Integrated Vehicle Health Management System on an aircraft is able to sense what's happening both inside and outside of the aircraft. Some aircraft can generate terabytes of data, every mission, which is vast. That data can feed the AI so that it can generate insights and recommend interventions allowing the humans to manage the maintenance regime, reducing costs whilst maintaining high aircraft availability.
“These sorts of technologies can be designed into the next generation of fighter jet so that more information is collected and processed. From there more insights will be generated and these will have tremendous value for industry and the customer.”
From raw data to useful info
Alex’s role as an in-house manufacturing data scientist is to look for opportunities. “How can we turn raw data into useful information that people actually want to see? Answering the ‘So what?’ challenge, so for example, what does this mean for the next mission, that's part of the interesting challenge.”
And part of the answer is the ability to pivot the manufacturing focus more quickly. “Traditionally, especially with high complexity builds like fighter jets, it's very difficult to change the plan. Gone are the days of the Spitfire, today’s factories are so specialised and optimised they have become quite brittle. That is something we want to move away from.
“We want to apply intelligent manufacturing processes so we’re able to adapt quickly, with the shop floor staff very much part of the picture. Everything — human and machine —supported by factory level intelligence.”
Enter the cobots
Alex sees a future that contains phenomenal possibilities. “In the near term we're looking at things like predictive maintenance and automatic inspection.
“Longer term, we will move towards software taking more of a physical hand, in the shape of robotics. It is not something we've done before because, as quite a low volume, as a manufacturer it didn’t make economic sense. But we’re moving towards intelligent robotics and these can understand the tasks required and work out for themselves what they need to do.
“We already have some applications of human and robotics working together — cobotics. But in the long term, the robotic system could become intelligent enough to infer preferences from the human co-worker. It may see a human struggling to hold something up and move in to help automatically. That's the level of reactive programming we may get to in time. That's incredibly exciting to me.”
Nick points out that this human–machine collaboration could also herald a new way of remote working too. He calls it telepresence robotics. “In the next decade or so, you will put your VR headset on at home and be immersed in the factory environment with the robot, and you’ll be able to take direct control.”
The impact of this shift is not confined to the factory floor. Think things like smaller car parks, no need for lighting to be on, and fewer toilets, because the bulk of the humans are off premises, but still able to do their job.
“AI and the IoT have the potential to change how we work, and also where we work. It’s an intriguing possibility.”
Designing the future
Applying these technologies as they evolve opens up creative possibilities for manufacturing. Both men can envisage a scenario where there’s a need for an aircraft that will carry out a specific role. The technology could recommend a form factor and break that down into build instructions and send it off to the factory. But that’s way over the next horizon. In the nearer term, once there’s a more flexible factory it provides a foundation for simpler things, like unmanned vehicles, to be produced quickly.
A whole new ... skills shortage
Alex is not unique in being involved in data science and data management within the company but few of his peers are specific to manufacturing. His route to his role is unusual too. He joined the company as a technical apprentice working on Typhoon.
“I've gone from making automated machinery towards making autonomous machinery,” he says.
Clearly in the new environment Alex is helping shape the role of AI which is going to be hugely important. But it’s also a challenging, complex arena and people with the required skills are currently in short supply. That’s leading some in search of jobs to demand rockstar salaries for their talents.
But Nick and Alex represent part of the possible solution to the issue from the company’s perspective. Says Nick: “There’s a global skills shortage for data scientists and AI engineers, and a severe shortage of people that can apply AI to the physical world. These are experts in autonomy [the embodiment of AI in the real physical world].
“There are great challenges when applying AI to the real world, where the consequences of getting it wrong are serious.”
BAE Systems has set up a Master’s programme with Cranfield University, with the blessing of the Government's Office for Artificial Intelligence. Alex and Nick are part of an 11-strong first cohort that’s helping to ensure the business has got the right skills in this high-demand area.
“We are competing for these people with the likes of Google so we need to upskill our existing staff and recognise that it may be a leaky pipeline. However, most of us are in defence, because we enjoy it. You’re involved in lots of interesting activities that money can't buy, learning about all kinds of applied defence technologies and these really spark people's interest and imagination.”
Alex is clearly one of those people. “For me the only better time to get into AI would have been five years ago. This is not just one of the great inventions, but in a sense one of mankind's last inventions. In that future, major inventions will all be done using AI. It's fascinating and to be very honest it's cool.”
The threat of the terminators
In the popular media there are dystopian arguments about AI but, says Nick, the vast majority of them are exceptionally unlikely. “They make an assumption that machines will have general intelligence like humans. That’s many, many decades away and we don’t have to worry about that kind of thing for quite a long time.”
Alex agrees: “People have concerns about AI but I would argue that by and large, they aren't the right ones. Media stories about AI typically feature a picture of the Terminator. That's not where the danger is.”
Instead both men raise the concern of flawed decision-making from using biased data and unethical applications.
“We need to worry about the ethics, regulatory and social aspects of AI,” says Nick. “At Cranfield we’re getting engineers to think about what they were doing, why they were doing it and to understand the regulatory and legal aspects. We want to embed good ethical practices within the AI skill set
“For example, there have been problems with bias in AI algorithms. That’s largely because the datasets being used weren't the most appropriate. The problem was not the AI, but the implementation of it. We've got to get all those things into the minds of our engineers to make sure that we absolutely get this right.”
Alex points to benign uses of AI like Netflix generating content suggestions for users and AI generating convincing passages of text as being more of a worry.
“There’s a potential here to create absolutely bespoke propaganda that hits the personal buttons of people based on what they will respond to. I don't think our society is in a good position to deal with that. But it’s probably a far more dangerous scenario to us than killer robots at this point.
This, they say, is why BAE Systems is focussed on developing the skill sets to apply AI ethically — and create great products.