The Davos Special

"Elder care is going to be the real draw for robots"

Rethink Robotics' founder Rodney Brooks believes that the best is yet to come for the robotics industry

Rodney Brooks can’t help but chuckle when you point out Hollywood’s obsession with robots running amok. To begin with, unlike in movies, robots in real life cannot write or rewrite the codes that can bring them on par with or make them superior to mankind. But they can definitely be made smarter and work more efficiently for humans and that is what the former MIT professor is engaged in with his start-up Rethink Robotics. With close to $100 million funding in its kitty already, Rethink is gaining in the popularity stakes thanks to its industrial production robot called Baxter. Brooks is no longer confining his research to industrial robots, having created a new version of bots that can be safely deployed in research labs. While he does see robotics getting up, close and personal with artificial intelligence, he is critical about Honda’s humanoid robot Asimo. In an interaction with Outlook Business, the 60-year-old indulges in a bit of crystal-gazing for the robotics industry.

How has the robotics industry evolved in recent years? 

Robotics research has been going on in universities in Europe, Japan and the US over the past 30 years or so. In the US, a lot of it is being funded by Darpa (defense advanced research projects agency) and the National Science Foundation (creator of the original internet under the NSFNET programme). The research is aimed at finding ways for robots to be deployed beyond traditional automation. 

I believe that sustained research funding coupled with the effect of consumer electronics products such as smartphones, which have made cameras and global positioning systems (GPS) really cheap, have helped the cause of robotics. Over the past decade, we’ve had 50 years of Moore’s law, considering the quantum of computation we now have in an embedded, low-cost system to enable real-time sensing and 3D sensing. [Gordon E Moore, co-founder of Intel Corporation, had in 1965 authored an article that predicted that over the history of computing hardware, the number of transistors in a dense integrated circuit will double approximately every two years.] 

So, I think that it was a combination of research and hardware pushed by the phone business that made the situation just ideal for robotic development. I think had it not been for the consumer electronics industry, 12 million Roombas [vacuum cleaning robots from iRobot] would not have got sold so easily. In a way, it is sheer luck, or else I would have gone to my grave without robots becoming serious business.

So, the initial phase for robotics was led by automation in the manufacturing sector?

Yes, in a very inflexible way. Traditional industrial robots were developed in the 1960s, when computers and sensors were really expensive. The intention was not to have them changing or interacting with people. We just had six things [features] happening, which meant that the cost of putting the robots into an automobile factory involved an integration cost that was four times the capital cost. They [manufacturing players] had to follow the ritual year after year to pay off that investment.

At what level of business does it make economic sense to deploy robotics? According to a PwC report, the total cost of a robotics system is approximately three times the cost of the robot itself?

Traditional industrial robots have that problem. Our robot, Baxter, is deployed out of the box — it plugs into a wall socket, it comes with cameras and other sensors, and there is no programming, just training the robots.  The idea is that it should be as easy to use as a laptop — you take it out of the box and start using it. This is very different from the traditional industrial robots. 

Does robotics need a developer community like Android’s?

There is already the robot operating system (ROS), which came from a company called Willow Garage. It is now run by a foundation called Open Robotics Software Foundation (OSRF). Our software is built on top of the ROS. With the research version of our robot, we have made a software development kit (SDK). Over the next 12 months, all that software will be under that SDK. So, anyone who has one of our robots will be able to put in new modules. We are going to sell not exactly apps but modules to improve performance. So, we are trying to develop our own platform. In a way, we are trying to build an ecosystem around robotics.

Is robotics finding applications beyond manufacturing?

The past two years have been a testing ground of sorts for this new category of robotics called Collaborative. The manufacturing industry, particularly in the US, is extremely risk-averse when it comes to new and unproven technologies. Yet, the early adopters continue to deploy these robots successfully and as they’re realising the benefits are indeed shaping up as advertised, they’re coming back and buying in larger volumes. We have the makings of a very solid snowball right now that continues to gain speed and mass as it rolls downhill.

Just look at the number of industrial robot suppliers. Initially dismissive of the category, they are coming out with their own versions of robots they call “collaborative” and you’ll get a sense of the market potential that’s just being recognised on a larger scale. We’re also seeing a great rush by many of those companies to catch up quickly when it comes to incorporating artificial intelligence into their machines. There’s a lot of money being spent right now to make up ground with the goal of being able to label their robots as “intelligent” as well. Either way, this category is taking off, and I expect the number of sold and deployed collaborative robots to go up exponentially over the next several years. They will be used more and more by small enterprises — our smallest customer is a nine-person plastics factory in the US. They will be used where the objects to be manipulated are more than three or four kilograms. 

Till recently, research labs did not have robots with arms, as the only ones with arms for sale were industrial robots, which are not safe. Today, our robots are engaged in 200 research labs around the world. In fact, a lot of those labs are working on robotic applications aimed at taking care of the elderly or even for carrying out routine domestic chores. I personally think that in Europe, North America and, in about 20 years, in China, elder care is going to be a real draw for robots.

How different will be the robot for elderly care vis-à-vis industrial robots?

I think it will be more like a tablet or smartphone with the person concerned using it. For instance, my mother is no longer able to get in and out of bed on her own. She has to wait for two people to help her do that. In this case, she has no control over the timing [of a person entering the room]. I would prefer having a robotic machine that could be told, “Put me into the bed now” and if I am not able to walk, have it do the manipulation on me, rather than relying on strangers to help me out.

Is that something that Rethink Robotics is working on?

We put Baxter out there as a research platform for anyone interested to work on. You need thousands of people to work on robots — only then will it translate into one or two good ideas. I want people to come up with inventions and they need not necessarily work on my robots.

What is your view on humanoids such as Honda’s Asimo?

The Honda robot is a marketing tool. I am sorry but it is not a practical robot. When the Fukushima nuclear disaster occurred in Japan, the Honda robot didn’t do its job. Instead, it was robots from my earlier company iRobot that went to Fukushima because they were practical and were already being used for bomb disposal in Afghanistan and Iraq. They were real working robots. The Honda robot is a show robot… it is not real.

Are we seeing hype being built around robotics, funneled by equally eager investors?

I hope it is not a bubble. Venture capital investments in robotics in the US have gone up pretty dramatically over the past three years. But I don’t think VCs are getting carried away on this front yet. The quantum of money being pumped into robotics is still minuscule compared with what is finding its way to social media. The reason for that is: successful social media companies with very little investment capital have changed the game. They are not delivering physical hardware, which is why the margin in their business is quite a lot. The initial capital delivers you quantum returns. But, in my view, social media companies are all competing with each other for the same set of advertisers. And I feel that there are no more advertising dollars left to get there. 

What do you make of the growing concern over robots taking away manufacturing jobs?

In the US, when farming got mechanised over a 100 years ago, the number of jobs in farming reduced dramatically. Except for farm-owners, these days, the only farming jobs are being done by immigrants. Though the number of farm jobs went down drastically, there were jobs at John Deere and Caterpillar. So, while the number of people in the farming sector shrunk, other parts of the economy automatically picked them up. I think the open question is, will that happen today? And the thing I pointed out in my session was that the side-effect of the mechanisation of farms was that the population of horses reduced dramatically. The horses didn’t find new jobs; no one plied them anymore. The question to be asked is: in the new context, are humans the horses? The answer to that is still not very clear.

Having said that, people, too, don’t want to do manufacturing jobs anymore. No one has got fired because of Baxter. People are aspiring for something better. Jobs in the West that people happily did 200 years ago are no longer in demand. Even in China, I spoke to many local manufacturers and their biggest problem was recruitment and retention. They just can’t get enough workers. 

Is it because they are being paid comparatively lower wages?

Wages are going up 15% a year [in China]. Actually, wages in China are now higher than Mexico. It is cheaper to manufacture in Mexico. What has happened is that essentially there is a middle class that doesn’t want those [manufacturing] jobs as they are really boring and repetitive. A lot of the repetitive jobs in the US were done by undocumented aliens [illegal immigrants]. The US policy has been to get rid of those undocumented aliens and now there is a labour shortage because people who were documented have much better jobs. But there is no denying that there is a disruption.

So, can human workers co-exist with robots?

You don’t really have to take a course in order to use a smartphone. It is built in such a manner that even an eight-year-old can figure out how to use it. You don’t need to be highly educated to operate a phone or download apps. Similarly, home appliances such as a refrigerator don’t need to be programmed, they work on its own.

When we started Rethink Robotics, I wasn’t thinking that the robot would necessarily be a humanoid. I knew it would have one or more arms (I liked the idea of three, just because no other robot has three) and cameras so that it could detect people and the objects that it was to interact with. And I knew it had to be intuitive to train so that factory workers would view it as a tool that they could use rather than some opaque technology that just made their workplace more alien. Over time, we decided that a humanoid form was the right thing so that the way the robot does tasks can be taken over by a person should the robot be busy elsewhere. This was also so that people would instantly understand what the robot was doing and through its gaze direction figure out just where it intended to move next.

So, I think that building machines that facilitate interactions between human beings will make people’s jobs far more interesting. So, there will be interesting jobs in the future that will be filled but this requires a lot of thinking about machines that people are already interacting with. That is what we tried to do with Baxter. You don’t have to be trained to use Baxter. A regular factory worker can figure it out pretty easily. So, we are trying to bring the aesthetics of the smartphone to industrial equipment.

There is the other aspect of robots going crazy. Is that concern for real or is it just paranoia?

Let me try and put this in context. These theories come from Hollywood. Hollywood tells good stories and entertains you. But the way they do it is that they have the current world and then they change one thing within that — suddenly, there is an intelligent robot injected with something or suddenly there are alien spaceships or supernatural forces at work. 

That is not what happens in real life; change happens gradually. You don’t suddenly end up having an amazing new device and struggle to figure out how to regulate it — nearly every step in advancement is regulated. In the US, refrigerators don’t catch fire and burn down houses because there are so many safety regulations that have been built up over time. Same will be the case with robots.

Companies such as Caterpillar and John Deere are looking at robotics in a big way. Where does that leave players such as Rethink Robotics? 

Those companies have roboticised a lot of their machines. Combine harvesters and a lot of the tractors today use advanced GPS and data from satellite imagery from different parts of the field and calculate how much fertiliser is needed based on that data. So, I don’t need to produce an automated tractor — Deere and Caterpillar are doing that. So, start-ups are finding things that people aren’t working on yet. But the challenge is that whenever a start-up tastes success, everyone tries to copy that.

Will the future be more about social bots than industrial robots?

If you had told people 30 years ago that over the next 20-25 years, your home will be full of computers, they would have probably replied, “what the hell are you talking about?” Because, back then, there were no microprocessors in households. Today, houses are full of microprocessors — just about every appliance has them. Similarly, as we talk about having robots in the house, we are still led by the Hollywood version of the robot. I think we will have robotic technology in the house and lots of it. But, just like the computers in the house, which don’t look like the computers of yesteryears, the robots of the future will also not look like the ones today.