Interview: iRobot CTO Paolo Pirjanian Talks Present and Future of Robotics

When iRobot acquired Evolution Robotics (the company behind the Mint cleaning bot) just over a year ago, Evolution CEO Paolo Pirjanian was brought on as iRobot's new CTO. Basically, this means that Paolo's job is to come up with cool new stuff for robots to do, and cool new ways for them to do it.

We got a chance to ask Paolo a few questions at RoboBusiness last month, and we sat down with him and Matthew Lloyd (iRobot's director of communications) to talk about the future of robotics and where he sees iRobot going from here.

IEEE Spectrum: The most exciting thing for us, when we heard that iRobot was acquiring Evolution Robotics, was the idea of localization and navigation. Can you talk about how those technologies might influence the direction in which iRobot is heading?

Paolo Pirjanian: You’re right, navigation is absolutely something that is very important to us and our firm belief is that navigation has reached a level of maturity now that is going to be influencing a lot of our products. We already see it: you have AVA at the high end, and you have Braava at the consumer level, and so navigation is the next wave of capabilities that we are adding to our product portfolio. Specifically, I cannot talk about roadmaps of products, but that’s for sure an area that we are focused on. And we do consider ourselves the leader. With the combination of the high-end navigation technology from iRobot, and the consumer-grade navigation technologies [Evolution Robotics] brought to the table, we are on the leading edge of technology, and these are mature technologies now that are going into products.

We talked to Nancy Dussault Smith a few years ago, right after the Neato XV-11 came out. I was asking her about the technology there: the Neato is a robot that can navigate around rooms in straight lines, and she said "we don’t want to do that because it’s more effective cleaning if you have a random pattern of multiple coverage." How does that reconcile with the idea of navigation being the future, and intelligence being the future?

So, there’s floor care and there’s things beyond floor care. For floor care, Consumer Reports actually just released a test they did, and Roomba, in terms of cleaning, definitely beats everyone. And that’s what iRobot has prided itself in about Roomba: Roomba gets the job done. iAdapt, which is the current diffusion-based technology from iRobot in Roomba, has certain advantages that allow it to do a good job of cleaning, and the evolution of navigation can complement that if combined properly. So we can improve the performance of the robot and speed up coverage and size of the areas we can clean with the same battery usage without compromising cleaning quality. And if you just look at the reports, the reports show that the LG, Samsung, and Neato robots that are doing systematic coverage are getting the task finished faster, but the cleaning quality suffers significantly, and we are not going to let that happen.

That’s something that other companies seem to focus on a lot: speed and efficiency.

This behavior may have been caused by the development of standards for robotic vacuum cleaners. One of the metrics that is used for measuring performance is area covered as a function of time, and so they are optimizing that metric, but as a result, compromising cleaning. And we think cleaning is the key function of the product, and we cannot compromise that. But, at the same time, we believe that, with a combination of the technologies of the two companies, we can keep improving our cleaning technology while improving all the other parameters as well.

As far as cleaning technology goes, how is combing these two companies changing that, since the way that each robot cleans is so different?

Braava's cleaning is complementary to both Roomba and Scooba because it deals with the fine dust and dirt that’s left behind. I would say, from a mechanical, suction, pick-up perspective, iRobot is the leader; we didn't bring anything to the table. The cloth-based mopping and sweeping is something that is complementary and I think it’s a good fit, that way.

When you say that they’re complementary, in your ideal robot-filled world, do you then see people having a Roomba and a Scooba and a Braava all working together in different areas of the house at different times?

We see a lot of customers that buy one product and then buy the second product that they feel is complementary. Usually, the third one is less likely, because Mint [he means Braava!] does both mopping and sweeping, so it depends on the need. Scooba does more like scrubbing, so if someone needs something deeper with dirtier floors, they will probably go with something like a Roomba and Scooba combination. Others that have a lot of hard floors in their home that require day to day maintenance will probably go with Roomba and Mint. And the majority go with one or the other, not the combination.

So do you see the future of these home care robots as everything converging on one robot that’s going to sweep and mop and vacuum? Or is it going to be these teams of smaller, more specialized robots working together instead?

I think that it’s hard to make that prediction, but one thing that we are very much focused on is that we want to make floor cleaning be a seamless task that happens in the background. The analogy I use is the sprinkler system for your lawn: you set it up and all you care about is a green lawn. I never worry about my sprinkler system unless I see a spot that’s getting yellow, and I think floor care is going to go there. We are getting, every year, one step closer to that dream.

Part of the reason I was asking is that I think the latest Roomba, or one of the more recent ones, had the wireless command center. Not something line-of-sight, but actual wireless, so isn't there a lot of potential there for communication between robots and, say, your phone, or other robots?

Right, so this, again, I think it could go either way. It’s hard to speculate whether it could be one robot combined with all of the functionality to clean any kind of floor, or whether you will have a combination of different robots. My prediction would be to start at single purpose robots and then gradually, later on, you’ll start combining functions into the same robot, when efficiencies of cost and technology are maturing. So you’ll probably end up with a little bit of both, that'd be my guess.

A lot of companies are starting to steer away from the term "robot." Are you focused on selling clean floors, or are you willing to say "we’re going to sell you a robot?"

Matt Lloyd: When we first came out with the Roomba, we didn't call it a robot. Because there was a bit of trepidation or intimidation. But the media, the consumers, took it back to "robot" for us. 

Paolo: Yeah. And that’s funny because when we launched Mint, I purposefully avoided the use of the word "robot" in any of our marketing materials. But now, for iRobot, our core identity is that we are the robot company. Consumers understand that, and I think there is a value associated with that for them to know that this is what we do and live and breathe every day is robots. We understand robotics. We may not be a Samsung or a company of that scale, but we are 100 percent focused on robotics.

But when you go down that road, then you're competing with this perception that people have of robots that they get from science fiction and popular culture. Roombas aren't like that. Is that an issue that comes up? That people expect them to be able to do things that they can't do?

Paolo: No, I think we have changed the perception in the market. People do look at Roombas and say "those are robots." Robotic floor cleaners. People understand the notion of single purpose robots now. So we have helped shaped the perception also, to ground it into reality.

Matt: Eighty percent of our customers name their products.

And what percentage dress them up in outfits? This happens all the time, right? People developing emotional attachments to their robots? Are you past that since you work from more of a technical perspective, or do you identify with robots in that way a little bit?

Generally speaking, in technology development and product development at iRobot, we are very much aware of that and there are subtle things we do to the products to also evoke that emotional attachment, which is very important. I think you will see more of that as we learn about more about what are the things that people are associating with. For instance, with Braava, we spent a lot of time fine-tuning subtle behaviors of the robot to express the internal workings of the robot, and when we put it in front of people, we saw how people refer to the behaviors: "Oh, it’s thinking," or "Oh, it’s doing this." We felt like we accomplished that, and that was very important.

Can you give an example of something in the Roomba’s design that was designed to evoke that sort of thing, and where you might go to increase to emotional connection that people get with their robots?

I’ll use an example that’s very familiar to me: when Braava sweeps the open area in a room, it has a built-in map, and it uses the map to go around the edges of the room, navigation around every single chair leg and so on. The way it's moving around to do that conveys an intent to the user, and people get it. You see the robot driving, and when it starts driving to get close to the chair leg, it will actually speed up, so it’s showing very purposeful motion: it’s faster than when it’s just sweeping. People say, "It’s trying to go somewhere to do something." When it approaches a chair leg—say the chair leg is over here—it comes to the chair leg, turns around, and "looks" at it. It doesn't really look at it, but people say it’s looking at it. It pauses, and then it starts sweeping around it. With Roomba, some of the jingles when you press the button, and when it’s done, have certain connotations to people. People say, "It’s excited to go out there and clean." And when it’s done, it’s a sense of accomplishment: "My Roomba is done!" So there are a lot of subtle cues that we use to evoke that emotional bonding.

Do Roombas still have a female voice?

The error five? [Laughs.] I don’t remember. I don’t think we have it anymore, but we can check.

I was just wondering, when you made that decision to give the robot a voice. It sort of forces people to think of it in a certain way.

Paolo: It changes expectations...

Matt: There’s a story that Colin [iRobot CEO Colin Angle] tells about when we were first focus-testing the AVA 500 [in hospitals], and we did it without the skin, and it was very much like a C-3PO. The response we got from patients was, "It moves too quickly, this robot is intimidating, it’s not doing what it’s supposed to do." When we put the skins on and we made it nice and clean, they actually accepted it into the room; they saw it as not being as threatening. None of the behaviors changed, but the way that patients were actually perceiving the robot had changed, all because of the way the design was built in to the robot.

How much attention do you guys pay to what’s going on in the research field? Is that something that you look at it, or is it usually too preliminary to have much of an effect on what you're thinking about and working on?

We have a pretty good sense of what goes on in research, and have pretty good ties with key research labs and people. And there are some areas that are strategically important to us, and in those areas we pay extra, closer attention to. I would say manipulation is an area where we are looking at very closely. And we are encouraging people to do research in manipulation the way navigation was being researched ten, fifteen, twenty years ago. We are also very interested in seeing any work that can have a profound effect on practical navigation.

When you're saying manipulation and navigation, it seems like PackBot is a platform that could benefit from advances there, especially with more autonomy.

Paolo: In defense settings, we don't want the robot to be in the position of decision making, and the warfighters are the ones that are making every decision. We are looking at autonomy for manipulation: simpler servoing, simpler commands, and so on, so they don't have to control every single joint in order to grab an object, but that's to the extent we have got on that front.

Matt: About six months ago, we announced a user-assisted payload that allows for maintaining one course of action. And if the robot hits a spot where it can no longer get radio signal, it will go back to the spot where radio signal was found. The only other thing it will do autonomously at this point is it will self-right, because what we don't want is for the robot to be down field in an area of high conflict, be flipped over, and then to have to have somebody go out there to right it. So those are three features that are moving towards that autonomous nature, but right now, that's as far as we've gone.

And why is that? There are a lot of other defense robot companies that are heavily focused on autonomy.

We do what our customers need in those markets, and our customers in those markets want full control. At some point they will come to us and probably ask us to add autonomous capabilities in there. One of the characteristics of iRobot is our ability to find the right business models and make profitable businesses out of very challenging technologies, and one of the ways that we do that is by listening to the market and customers very carefully. So, yes, I know there is a lot of work going on on autonomous UGVs and so on, but the customer has to be ready for it.

How much overlap is there between the research you do with defense robotics and home robotics?

In terms of autonomous navigation, there is a lot of overlap. There are a lot of technologies that we are developing for AVA, like visual navigation, that will benefit both AVA and our home business. And with some work those technologies can be adapted for our defense business. We try to look for synergies as much as we possibly can. In other areas, like the hardware that you have for outdoor defense robots, this thing has to be able to take a 15 meter drop onto concrete and survive, and we can't afford to have that kind of ruggedness on a consumer product, right? Because it adds costs.

Yeah, I haven’t tried that with my Roomba!

The Roomba probably will not do well, but you may be surprised. I won’t encourage you to do it, though! I actually tested one of the first Mints out of production: I went to the factory and said, "Let’s go outside and test it." We were walking down a flight of like 20 stairs, above a marble floor, and I just threw the robot down. And the factory guys went "Oh my God!" But the robot didn't even get a scratch.

Is that the normal testing that you do?

That’s the CEO coming to visit! [Laughs.]

I guess that’s one of the things that’s hardest to do with home robots, to make sure they don’t break. It should be a lot of work. It’s definitely exciting that you’re coming out with a new Roomba, but I hope that this is not the end of the line for Braava, either.

We are committed to Braava. We're expanding it!


Many thanks to both Dr. Paolo Pirjanian and Matthew Lloyd for speaking with us.

[ iRobot ]

Related Stories

Automaton

IEEE Spectrum's award-winning robotics blog, featuring news, articles, and videos on robots, humanoids, automation, artificial intelligence, and more.
Contact us:  e.guizzo@ieee.org

Editor
Erico Guizzo
New York, N.Y.
Senior Writer
Evan Ackerman
Berkeley, Calif.
 
Contributor
Jason Falconer
Canada
Contributor
Angelica Lim
Tokyo, Japan
 

Newsletter Sign Up

Sign up for the Automaton newsletter and get biweekly updates about robotics, automation, and AI, all delivered directly to your inbox.

Advertisement
Advertisement