In this exclusive interview, Dr Andrew Goldenberg, founder and chief technology officer of Engineering Services Inc, and currently also the chief technology officer of publicly listed Super Robotics Ltd of Hong Kong and of Anzer Intelligent Engineering of Shenzhen, China offers his perspective
When you’ve been in the robotics and automation industry for more than 35 years, you tend to form opinions that other people want to hear and learn from – even if those opinions are not entirely complimentary to the listener.
Dr Andrew Goldenberg is professor emeritus at the University of Toronto and chief technology officer at Super Robotics, Engineering Services, and Anzer Intelligent Engineering – all part of the same group – and he’s an avid reader of Robotics and Automation News.
But just because he likes many things about the website doesn’t mean he agrees with some of our more “offbeat” articles, as we call them – the possibility of alien robots from Planet X infiltrating life on Earth and so on.
“I’m ready to call it stupid,” he says in an exclusive interview with Robotics and Automation News.
It’s not exactly a comment we can use in our marketing literature, but glad he feels comfortable enough to share his thoughts. Now, on with the interview.
Yeah, well, that’s just, like, your opinion, professor
Goldenberg’s observations and opinions lead us to believe that we are working too hard and publishing too many stories, some of which may not be worthy of the seriously advanced technology and serious technologists we are fortunate enough to have as readers.
He qualifies his criticism by saying he finds the media in general hypes things up too much, so, in that sense, Robotics and Automation News is no different.
He suggests that more editing is required – and less content, a suggestion which may correlate our tendency to publish too many articles with prolific academic publishing, the wider economy, and the general well-being of the common man and woman in the street.
“Who says that publishing more is good for society as a whole? Who decided this?
“This is actually not the case.
“Stop bombasting about this and that as a way for growth. I claim it is not.
“It serves those who are lucky enough to be educated and be employed – I am one of them, but I know that many people are not.
“So, this excitement with the competition on the number of articles, for example. I’m ready to call it stupid.”
Goldenberg is clearly passionate about this subject, as he is about many others, most of all perhaps the idea that the field of robotics and automation should be open to everyone who has the aptitude – regardless of whether they have a PhD or not.
He says there has traditionally been an over-emphasis on highly educated “super-intelligent” individuals.
“I want to make robotics a field where ordinary people – not only the super-highly trained – could find employment and that’s something I feel is very important.
“I’m not in a position to offer solutions but I raise that because we do need super-intelligent people but, socially speaking, not everyone can either financially or mentally become a PhD or super-inventor.
“There is a bit of over-emphasis on the educational background of people that are needed for making progress in the robotics and automation domains.
“I think, as a union of people dealing with robotics, we should make this field amenable so that we can grow like the car industry, which has become a super source of employment for many people with or without high degrees.
“I’m not sure that’s possible. But I think robotics will become as large an industry as automotive, so I would rather lean towards making this industrial domain affordable – not only in terms of affordable robots but also in terms of affordable employment.”
This opinion – the idea that the robotics industry will become as big as the automotive industry – is one that is shared by many people. In fact, the way things are going, automobiles themselves are becoming increasingly robotic.
The market is everything
One of the key observations Goldenberg makes about the robotics and automation sector worldwide is that the vast majority of activity is in the field of academic research, rather than in commercial or industrial settings.
He has written about this before in IEEE publications, saying that robotics evolved as a novel academic activity, which is fine, but that’s where it seems to have largely stayed.
Goldenberg suggests learning from the computer industry as well as the automotive industry. They, too, started as fringe or marginal activities, almost as trivial pursuits. But both are now among the largest industries in the world.
He predicts a similar trajectory of development for the robotics and automation sector, but says there is a long way to go.
“Robotics, as a field, is as old as computers, but lags behind in terms of establishing its industrial business presence,” says Goldenberg.
“That does not negate successes. There are companies that are hugely successful in certain domains, mainly industrial robotics – Kuka in Germany, Fanuc in Japan, ABB and so on.
“If I look at – and this is my own guesstimate, not really scientific or precise… If you ask me, ‘How many people deal with robotics across academia, research institutes, companies, and so on… how many people deal with real business in an industrial sense and how many deal with technology development and new ideas?’
“I would say that it is disproportionately low on the industrial or business side.
“I would say 25 per cent at best does business in robotics, 75 per cent does research and development.
“But the robotics field is expanding. Now, it includes cars with autonomous navigation and even drones. Some people think that drones belong in robotics, and you can argue that’s the case.
“There is also an impetus to encourage startups with new ideas, but unfortunately the ideas do not always respond to a need, definitely not from the market.
“They tend to consume a lot of brainpower from the most intelligent people but I don’t see the business possibilities. A lot of startups are just startups.
“The fact that a website like yours lavishes them with praise because they raised funds is not proof of success. Raising funds is the ability of the brokers to sell shares in the company to raise funds – that’s it.
“Success is when you’ve got a product, you sell it, and you make a profit – nothing else.
“In robotics, that profit-motive approach is not as yet dominant. And that’s too bad.
“So we need to make some effort to highlight commercial success, to encourage profit-making machinery, because that’s what it is – it’s a business, as opposed to only technology, and innovativeness, and raising funds, and other things that come with it.”
Clarity of thinking about the commercial possibilities of the technology one is developing is essentially what Goldenberg is espousing.
Although Goldenberg himself is from the academic world, he founded what is now a highly market-oriented company in Engineering Services Inc, or ESI.
The company’s approach has attracted investors from China, the world’s largest market for industrial robots, and now, ESI is part of the Super Robotics group, with a business unit called Shenzhen Anzer Intelligent Engineering in Hong Kong.
This Chinese connection has opened up far more opportunities in what Goldenberg says is ESI’s main revenue-generating activity – the development and integration of industrial automation systems and solutions.
ESI has also released a new version of its collaborative robot, the C-15, thereby entering the fastest-growing segment within the industrial robots market.
Also included in the company’s extensive product line-up are mobile robots, often used as security guards to patrol buildings and homes, or to deliver items. One of these can be used in explosive disposal and in other dangerous situations.
The company has been awarded many patents for its inventions across its product range.
In all, ESI has five product and service lines, including what Goldenberg calls “custom robotics”, by which he means industrial automation in a typical manufacturing plant, automating processes and so on.
But perhaps most crucially, given possible future trends, ESI has a range of service or personal robots, a sector which Goldenberg believes is next in line for growth in much the way the industrial robots market has experienced over the past decade or so.
“Short term, industrial robots are dominating and will probably still be dominating the markets,” says Goldenberg.
“Long term, personal robots.
“If you think about personal robots, you can think about personal computers – everyone or almost everyone has a personal computer, laptops of different sizes, maybe a tablet, a smartphone.
“And the next step on from that will be a robot.
“So the big market, the real market, is the mass use of personal or service robotics.”
Goldenberg credits robotic vacuum cleaners with providing an entry point into the consumer space for many robotics companies, and he credits one robot in particular.
“It’s effectively the Roomba, the robotic vacuum cleaner, which was a nice entry into that domain, making robots household items.
“Reception robots, telecommunication devices, and so on, will also be part of this market. We call these service or personal robots – we use those terms interchangeably, service or personal.
“This is what the market is heading into. And I think that in less than 10 years, maybe five years, you’re going to see robots around the house.
“There are attempts by startup companies producing various devices that may do some of this work, and not only in a household setting. It could be a public setting – malls, banks, or other places. There are robots already doing some work in restaurants.
“These are really nice attempts, but this is really not the core. We don’t know how this industry will evolve because it’s not only about the technology of moving around and not bumping into people and talking to people and smiling to people – that is the short term.
“There is the issue of security, control over the robots, effectively protecting them – and that is a major issue.
“So, I think five years is about the right amount of time to start seeing them safely in malls, doing the job of, for example, a portable information desk, if you want.
“At home, they could do anything from assisting elderly to people with physical disabilities around the house, and so on. That is the market.
“I think that when we reach that point, industrial robots, as used today in manufacturing, will be second to service robots in terms of volumes of sales.
“Then there are other domains. The mobile robots for delivery. Basically, you have drones delivering packages.
“Delivering is the easy part. It’s the infrastructure to enable them to do that safely and in a regulated way is tremendously complicated. I’m not sure how they are going to solve that.
“I also think autonomous cars with human-driven cars on the same roads at the same time will pose problems.
“But the industry is just looking at technology, asking questions like, ‘Can the car move without bumping into anything?’
“But that’s not the difficult part. The difficult part is that you have cars driven by people alongside autonomous cars, and I don’t know how they’re going to survive together.
“One day, maybe 30 years from now, there will be no private cars, everything will be done autonomously – that I understand.
“I don’t know how autonomous cars and human-driven cars are going to be on the same roads when faulty software, unreliable software, creates tremendous problems, and the whole idea is that we are just working towards a point where everything is safe – that’s nonsense.
“It is not safe.”
The lag in mechatronics
Many would agree with Goldenberg’s assertion that the mechanical aspect of mechatronics is somewhat lagging behind the other parts.
While software can be programmed to create virtual robots of any shape or size inside a computer, building a workable version of those machines is not always possible because the materials and mechanical components are not up to the job.
It’s probably the reason we haven’t seen humanoid robots which can move with anything approaching the skill of an actual human.
Goldenberg breaks it down. “There are really three sectors working together in robotics and automation, and mechatronics.
“One sector is the mechanical components. Some are electromechanical, but I’m thinking more in terms of the nuts and bolts and materials.
“Then the second sector is electronic, and here the chip is central – the size of the chip, the memory of the chip, and the data processing of the chip, and so on.
“And then there is the software.
“So, in the context of robotics, all three are needed, they are equally necessary.
“The development of each branch follows the route of developers or companies – some are more mechanically oriented, others are into electronics, and others are in software.
“Then there is the integrator, who makes the robot, putting all three fields together.
“So if I were to classify where things are, I would say software is way ahead. Electronics is second, not far behind. The lag is in mechanics.
“Mechanics means materials, gears, transmissions, and so on. And all of these are – in comparison to software and electronics – way behind, making applications difficult to integrate because of many reasons… weight, size, and so on.
“This is the way I see it.
“Software seems to run ahead, and software is, by nature, not connected to the real world because it’s basically digital.
“Electronics is digital partially, but also has to do with the physical things – integration of chips and so on – for which there are well-established suppliers.
“Mechanics is well behind – there’s no question about it. The first thing I think about is materials. We have aluminium and other materials that have not changed for decades.
“We are primarily in the mechanics area and in electronics, and then after that, software. So, in our company, we develop in the reverse order with respect to the advancements are out there.
“Mechatronics, by definition being mechanical electronics, means that mechanics and electronics are the most important.
“But nowadays, the software is not just for integrating the system. The software provides a lot of ways of compensating for the shortcomings of either the mechanics or the electronics.
“So, software is a fairly magic and important element of those systems. There are the people who work in software, there are a lot of advancements and ideas and so on, artificial intelligence is part of that software mix.”
While many – including us – believe that AI holds the promise of a future full of vastly superior robots to the ones we see now, Goldenberg says this is not an accurate perception for now, that we may be over-estimating the power of AI in the robotics area at the moment.
“There’s a lot of hot air about AI and how it’s going to influence robotics,” says Goldenberg.
“AI works extremely well in data mining, e-commerce and so on, but it has severe shortcomings for electromechanical systems like robots.
“We are trying to figure out how to do that. We are nowhere near a solution. But if you read the media, like your website, ‘It’s all over, it’s all there, AI will solve everything’.
“We need more people to do AI – that is correct. And of course there will be many advances.
“But there is AI for robotics which is woefully complicated, and there is AI for things like e-commerce that is different – that is doable, it is being done, and it is successful.
“So we have a bit of discontinuity in excitement. I get excited too, but sometimes I feel disappointed that there is too many promises and expectations – they are way too high. We may get there, but let’s take one step at a time.”
Goldenberg makes the point that while AI is good for digital-only processes, it is yet to be developed properly for mechanical systems, and it may take a long time before AI significantly changes or improves robotics and automation systems.
Personnel working in the ESI’s five robotics sectors are looking into how AI could be applied.
Research is part of the company’s heritage. “The company originally, before it was acquired, was primarily a developer of technologies so the emphasis was not necessarily on product sales,” says Goldenberg.
“After the acquisition, the focus of the company’s activities has shifted somewhat from technology research and developmental work to products.
“The technology, which includes patents, is hugely successful. We have patents in all our five areas. And we have in-house technology that we currently embed in products.”
The Da Vinci code
As well as AI in robotics, another technology that Goldenberg says will take a long time to develop is the company’s medical robotic technology.
Some readers will have heard of the Da Vinci robotic surgery system, and similar systems emerging now, but what Goldenberg has in mind is something different, and quite radical.
ESI has robotic surgery systems that can operate inside a magnetic resonance imaging machine.
MRIs are large enough to contain a human, who is usually lying down and being slowly conveyed into the scanning area. Hospitals use MRIs to scan a patient’s entire body for tumours and other helth problems.
It is inside this machine that ESI is looking to place a robotic system, and it has already built test solutions.
But the development of a commercial version is fraught with many difficulties, as one might expect from what is probably the most stringently regulated business sector – health.
“We’ve been working on medical robotics for many years, on and off,” says Goldenberg. “It’s all in development. We have developed solutions for testing.
“We’re looking at things like prostate surgery inside the MRI with a robot, so the patient and the robot are inside the MRI because the MRI can provide a high-resolution image of the location of a tumour in the prostate so the prostate doesn’t have to be extracted.
“In normal practice, the prostate gets extracted.
“That’s one application.
“We have an application for some types of brain surgery, and we have other applications.
“But there is no surgical robot that does everything, although there is a famous robot called Da Vinci that was created by Intuitive Surgical.
“That company is hugely successful, has created a robot that is used in many hospitals around the world.
“It [Da Vinci] is a large robot, doesn’t work in MRI.
“Magnetic resonance imaging has always been used as a diagnostic tool, providing an image of tissue that may be subject to health problems.
“Now, the MRI also becomes a medium for performing surgery, and that is what we are working on.
“It’s a long road. It’s not something that happens quickly. It’s not just about having a robot available.
“We develop these robots with medical personnel, with doctors, together, so we’re not developing this by ourselves, we have medical experts involved.
“But even medical experts do not always agree on surgical procedures. One hospital does one thing and another hospital does the same thing totally differently.
“So, if you work with one hospital and develop a robot for their needs, it may not be accepted by the other hospitals – that happens even in the same city.
“So, medical robotics, I would say, is a challenging, interesting and quite difficult area in terms of business to justify the investment.
“Surgical robots have been shown to improve healthcare, but as a business person, I would say this is a long-term investment, not short-term.”
The time scale Goldenberg puts on a commercially available MRI robot is between five and 10 years, mostly because of all the regulatory red tape that a company like ESI would be legally required to go through.
But this is also potentially a huge market for the company, so development work will continue at ESI.
For now, though, the Romanian-born Goldenberg, who has been living in Canada for 46 years, is concentrating on what will make ESI a commercial success, especially in China.
“At this point, our focus is on the Chinese market,” says Goldenberg. “The Chinese market for robotics is the largest in the world at this point in time, so most of the focus of the company is on that market.
“We do obviously operate in the Canadian market, and in the US, but most of the emphasis on the business side is in China.”