Machine learning: a manufacturing risk or revolution?

Patrick Williams discusses the future of machine learning in manufacturing.

Patrick Williams discusses the future of machine learning in manufacturing.

CAD technology has the potential to revolutionise the speed and precision of engineering design, yet some believe it poses a threat to the wider industry workforce. Manufacturers’ Monthly reports from Autodesk University on both the possibilities and the implications.


Like most disruptive discoveries, the dawn of machine learning is double-edged; from the big dreamers who recognise its potential for limitless creation, to those daunted sceptics who fear its “destructive” power.

For now, computer-aided design (CAD) is still led by human direction.

Yet, the speed in which artificial intelligence changes the way the industrial world is engineered is picking up at such a rate that some believe the mortal face of manufacturing could disappear altogether.

The debate continues. Should scientific discovery endure even if it puts the livelihoods of traditional engineers at risk?

All the while, planes in the sky are becoming more streamline due to stronger yet lighter designs composed by split-second algorithms, and the places we live, work and travel are being made more connected through the introduction of augmented and virtual realities.

For Autodesk, this is a common question. Is the intuition of its workforce putting its very existence in the balance? Or, by reinventing the way things are designed and manufactured, can machine learning open doors to a world their predecessors could only imagine?

Patrick Williams, Autodesk’s senior vice president for Asia-Pacific, is excited by the possibilities.

“Intellectual production is changing and today it is common to work on teams that are not just local but sometimes global,” he told the annual Autodesk University conference in Sydney.

“Using technologies like augmented reality and virtual reality are going to allow us to change how we understand designs.

“How we predict and make decisions and, ultimately, how they are going to get to a better design in a more collaborative design environment.”

Optimising design

In the future, machine algorithms will completely automate the process – spanning design, engineering and fabrication – according to Williams.

In theory, this will offer optimal designs for anything from the composition of a lightweight chair to the infrastructure of an integrated city, all at the push of a button.

“I think that it is clear all the things we make – whether they are buildings, cars or highways – are all deeply connected to each other and do not function in isolation anymore,” Williams continued.

“They talk to each other and generate a tremendous amount of data and insights, which change over time. It is a combination of these things that, when combined with a third disruption, it creates a significant disruption for all of us.”

That third disruption is machine learning, which Williams says is accelerating at an “alarming rate”.

“It is changing what we design and make,” he said, “and also how each and every one of us works.

“There has been expressed concern that these types of technologies are going to take away jobs. At the same time, they are also enhancing the way we work and what we produce.

Engineers and architects attended Autodesk University in Sydney.
Engineers and architects attended Autodesk University in Sydney.

“Human and machine partnerships are now allowing us to create smarter and better products than ever before.”

The biggest disruption, according to Williams, is also a burning issue beyond industry.

“Climate change is forcing each and every one of us – and each and every industry that we design in – to think differently about what we design and how we design it,” he continued.

“Everyone has to ask the question of what we should do in a world where climate change and resource constraints are rampant.

“This is where sustainable design comes in. It’s about creating designs that are better for the world, for us today and for the children of tomorrow.”

Motivating the manufacturer

So, where does the manufacturer fit in?

The industry in Australia has been presented with a choice: whether to seek a cheaper production line abroad so to avoid rising labour costs at home; or alternatively to stay put and reconfigure the workforce to meet changing technology.

Eric Begeja, an executive advisor for the industry consultancy firm, GHD Australia, insists modern-day engineers need to embrace it.

“I am originally from a manufacturing background,” he said, “and the manufacturing sector, over the past 30 to 40 years, has gone through a lot of technological change.

“Labour costs are going up in Australia and there is cheaper labour in Asia, so they have to decide whether we close our facilities here in Australia or to stay back and automate.”

From an Australian perspective, he says parts of the industry have lost its income to the economy, while those who have stayed in Australia have reskilled and retrained a lot of their workforce.

“We have obviously lost a lot of jobs, but we have kept a lot of jobs too, by retraining the workforce,” Begeja said.

“As an engineering consultancy, we have got this dilemma: do we carry on the way we are, do we utilise low-cost sensors, or do we transform our workforce by utilising some of these newer technologies?

“I have to make a confession. I am also one of those poor engineers that is potentially facing extinction and it is really in my best interest to make sure I understand where that technology is going, so that I can adapt, evolve and keep my species going, too.”

In his opinion, there are two scenarios that could play out. Where computer technology is far more intuitive, the human engineer has to think smarter too, which means understanding his or her own strengths over artificial intelligence.

While, on one hand, machines work infinitely faster than any human can, understanding a problem and where to place most of your energy are key human traits that may never be replaced.

Another is the eye for a beautiful design, in addition to understanding efficiency.

“Do we embrace this change, resist it, or collaborate,” Begeja said.

“There is an inherent fear in all of us that means, when we are threatened, we go in to this defensive mode where we try to discredit or put down the technology that threatens us.

“If you think it is a threat, you should really know your enemy, which is important for all of us. If you can’t beat them [machines], then join them.”

Infinite ideas

In the past, the role of CAD used to be passive and, ultimately, owned a finite function.

When building a car seat for example, traditionally, an engineer would have had to run a simulation of a model, which could take hours before he or she received valid feedback on the design’s performance.

A further redesign would follow and, unless the car seat passes the next simulation, the cycle continued and, even then, didn’t necessarily mean an optimal design had been achieved.

Bupesh Lall, Autodesk’s Field marketing director for Asia-Pacific, discusses the opportunities smart computers can present the industry – especially coupled with the role of 3D printing.

“In a sense, what I was doing [as an engineer] was proposing and disposing designs to get something better,” Lall said.

“Now we can flip the whole thing around. Instead of trying designs one at a time, we can explore many different options with no physical risk or cost.

“The value of infrared computing is that is can extrapolate, mutate and mash up an infinite number of options for a design simultaneously, which is the world we call ‘generative design’.

“Using intelligent algorithms that are based, in some part, on evolutionary patterns of nature, it opens up the design space to many new solutions.”

In aerospace, Autodesk is working with Airbus around the cost efficiency of its planes by considering how different designs can shed the weight of a fuselage and, effectively, long-term fuel consumption.

“The relationship between us, the software and the tool used to be very directed [but] this relationship, courtesy of generative design, has evolved to be much more adaptive,” Lall explained.

“The generative design tools anticipate our design needs and, in doing so, have discovered the best designs that we may never have discovered alone.”