Robots, cobots and Cambridge’s role in the next industrial revolution
PUBLISHED: 00:00 31 December 2016 | UPDATED: 09:16 31 December 2016
From serving food to carrying out deliveries, harvesting crops to serving warehouses, robots can carry out an ever-growing range of jobs today. Following our feature on surgical robotics, PAUL BRACKLEY speaks to Chris Roberts at Cambridge Consultants about what’s coming in other fields.
In Colorado, a delivery of 50,000 cans of Budweiser was recently completed following a 125-mile trip. Unremarkable, you might think.
Except that the truck carrying the beer was driverless. Or rather, once the driver reached the interstate, he hit ‘Engage’ and let Uber’s $680million ‘Otto’ truck do the rest.
The world’s first autonomous truck delivery is a very clear sign of what’s to come.
Meanwhile, in Mountain View, California, you can order pizza made by robots working alongside humans. And Zume Pizza, a start-up, expanded on its innovation in September by adding smart ovens to its delivery trucks that come on at just the right distance from your home to finish cooking your pizza en route.
Robotics and automation have been with us for decades but the pace of development being seen today is remarkable. What’s changed?
“There’s been an evolution of the technology,” explains Chris Roberts, head of industrial robotics at Cambridge Consultants. “There hasn’t been a key breakthrough – everything has just got a little bit better. All that evolution has enabled some more revolutionary applications. Things that were science fiction 10 years ago or university research projects five years ago are now things you can imagine engineering and building in the real world.
“A very public example of that is self-driving cars – those are happening.
“We’ve done an automation project for Ocado, which is a warehouse automation – a lot less glamorous than self-driving cars but still something that could not have been done five years ago. There’s no fundamental, key enabling technology that’s happened – things like vision algorithms have got better, robotics off-the-shelf technology has got better.
“That’s opening up new sectors that haven’t really been automated before. Logistics is one example. Big companies like Ocado and Amazon are bringing more and more automation to warehouses.”
Cambridge Consultants were tasked with helping Ocado, the country’s largest, online-only grocery retailer, to create an autonomous warehouse. The firm created a wireless control system in the 5GHz Wi-Fi band that enables autonomous robots to move around above a grid, storing and retrieving crates stacked within it. Ocado can control 1,000 robots from a single base station, communicating with them 10 times a second in an area the size of an Olympic swimming pool.
Chris said: “We are not about getting a little bit extra performance out of a car production line. It’s about bringing robotics and automation to new and innovative areas so trying to automate things that haven’t traditionally been automated.
“Another area that’s really big for us is agriculture and precision agriculture. Farms have been automated to a certain extent for decades with combine harvesters, for example. But look at something like apple picking. There are something like 10 billion apples picked in the US each year and every one of them is picked by hand. That is one of the things that it’s now possible to do.”
Cambridge Consultants unveiled its apple-picking robot back in October 2015. It paved the way for robots to take on complex picking and sorting tasks involving irregular organic items, such as sorting fruit and vegetables, or locating and removing specific weeds among crops.
“It’s surprising how dumb most robotics is,” says Chris. “Most robots can cope with doing the same task over and over and they can do that task very well and very precisely but only if it’s exactly the same task. Something like picking apples, where the apples aren’t in the same place, move around a bit and don’t look exactly the same, is actually really hard for robots.”
Hard – but no longer impossible.
“There were two big things we needed to develop for the apple-picking robot – one was the vision algorithm so that we could work out where the fruit was and where to put the grippers, and then to work out what sort of fruit it was, as the grippers needed to be able to handle soft, natural products that were different sizes and shapes.
“It’s those sort of semi-structured tasks that we’re seeing more robotics in now.”
The agricultural robotics market is predicted to be worth $15billion by 2020. Imagine a farmer using an autonomous tractor, flying a drone to survey crops and using robots to weed, spray, prune and harvest crops.
“Harvesting corn or maize is automated already but it’s not precise – you harvest the whole plant and then mechanically sort the bit you want,” said Chris. “If you looking at high value crops like lettuces, broccolis, berries or fruit you can’t harvest in the way you do corn.
“A lettuce needs to be tipped to one side, cut underneath, handled quite carefully and there’s a reasonable amount of quality control done by the picker. So we generally have people walking the fields picking lettuces. Those are the sorts of problems that are engineering challenges now rather than research ones and if there’s somebody with sufficient money and motivation to invest in that you could imagine a more automated solution in the near future. It’s all about the business case: can I make something for enough money to get a return on my investment,” says Chris.
“Robots walking around being mistaken for humans is still science fiction – that’s still many years off. Things that we find easy can be much more difficult for robots. It’s much easier for a computer to play chess than walk upstairs.”
Robots, then, are typically good at a single task.
“Some months ago an artificial intelligence was built that could beat a human Go player – Go is much harder to analyse than chess is so that’s a really impressive accomplishment. But that AI can only play Go. It can’t walk down the street or open doors,” says Chris.
Cambridge’s very own Prof Stephen Hawking has warned of the dangers inherent in AI.
“There are lots of theories that once you’ve built an AI that is good enough to create its own AI that very rapidly you get singularities and evolution in a way but that still does feel like science fiction to me,” says Chris.
“I was watching Humans [the Channel 4 drama on robots] a couple of years ago and there was a robot that, apart from flashing green eyes, was indistinguishable from a human. And while that was on the telly, I was working in a lab with six others trying to get a robot hand to pick up an orange – and we just about managed to do it in six months without squashing the orange.
“That gives you an idea of the gap between science fiction and reality.
Skynet – the self-aware robotic force that declared war on mankind in the Terminator films – is still a long way off, then.
But humans and robot interaction is with us now. One of the biggest recent development in robotics is the rise of the cobot – or collaborative robot.
“Traditional robots that you find in a production line are very good at moving objects around,” says Chris. “They’ll pick up a car door and move it from one place to another. And they’ll do that whether or not there’s a human standing in the way, so you have to segregate them.
“With collaborative robots you don’t need to do that. If they sense a collision with a person they stop without hurting them or the motors are torque limited so that if they try and drive the arm where a person is, it stops. So these are thing you can buy – and crucially there are standards that you can certify your system against to prove it’s safe for people to interact with. Once you got that, there are loads of applications that you couldn’t do before because people can be in the same place and share tools.”
Research is now focusing on robots that don’t just stop but plan around a human.
“So you can imagine a robot working with a human in a lab and the human reaches for one test tube, a robot would reach for another. That kind of thing is still research,” says Chris.
With Amazon working in Cambridge on its ground-breaking delivery-by-drone system, Prime Air, and rapidly growing robotics applications across transport, logistics, agriculture and the service industry, it seems reasonable to say we’re heading for another industrial revolution.
The Bank of England’s chief economist Andy Haldane has warned that up to 15 million jobs in Britain are at risk of being lost to an age of robots. A third machine age will “hollow out” the labour market, he predicted, widening the gap between the rich-poor gap and threatening in particular those in administrative, clerical and production roles.
In an article for Newsweek last week, global markets’ consultant Amro Zakaria Abdu warned: “Governments around the world … will be forced to come up with new social programs to mitigate the inevitable tsunami of unemployment.”
Chris is more sanguine about the notion.
“People always fear robots taking our jobs and I think it’s true given these new applications that there will definitely be industries where there will be disruption.
“But robots and humans still have very different skill sets and robots are getting more capable but there will always be some tasks that cannot be done by a robot. The new tasks that robots are taking on are the more mechanical, less exciting jobs – there are not the jobs that people aspire to do, or want their children to do. And there is becoming a shortage of easy, manual labour.
“The evidence from places like Germany, which has a lot more industrial automation, is that it leads to the creation of jobs not just because you need people to support the robots but because generally your factory gets more efficient so you making more stuff but there are always some tasks in the process for which you need humans.”
Low-paid, entry jobs in particular industries seem to be those most at risk.
“in a semi-structured environment like dispensing ice cream, that’s the sort of thing that I could imagine robots doing,” says Chris. “I’ve seen an interesting robot that does room service deliveries within a hotel. You can build wireless sensors into the lift so it doesn’t have to press the button and it can knock on the door without having to physically doing it.”
McDonald’s is known to be working on using robotics in its fast-food restaurants although the company’s CEO Steve Easterbrook has been at pains to point out that as a service business it “will always have an important human element”.
If a Big Mac from a cobot doesn’t get your taste buds tingling, you could always try investing in Moley, dubbed the world’s first robotic kitchen and set to launch next year. Featuring a pair of fully articulated robotic hands, it will cook hundreds of dishes for you from an iTunes-style library.
Does this herald a widespread introduction of robotics into our household then?
“It’s hard to predict but I am less convinced that in the near-term future we’ll have lots of robots in the home,” suggests Chris. “The scale and the cost is one order of magnitude too much. A robot arm might cost £20,000 which is very cheap for an industrial process but not for the home.
“I’d love to see more automation in the house – robots that do ironing. I’d buy one – but not if it costs £20,000.”
So what is Chris working on now?
“I’m working on a very exciting automation project in an area that’s not traditionally automated…and that’s about all I can say.”
Watch this space.
Surgical robotics advance again
Cambridge Medical Robotics has unveiled its next-generation surgical robotics system Versius.
The system has performed upper gastro-intestinal, gynaecological, colorectal and renal surgery in trials on cadavers in October and November.
CMR says Versius can visualise and access all these surgical workspaces and perform tissue manipulation, suturing, needle driving and electro-surgery.
The company has now built 20 proprietary robotic arms, built and tested nine variants of its fully articulated 5mm instruments and conducted 11 usability studies. Some 32 surgeons have used Versius so far.
As reported in our first robotics feature, surgical robotics is advancing rapidly with smaller, more flexible and cost-effective robots being developed.
CMR technology director Luke Hares said: “This has validated the universal surgical robot concept that CMR has been developing – using numerous robotic arms in a flexible, modular, system to provide the versatility and dexterity necessary to support the majority of laparoscopic procedures.”