People love robots. If you’re here, you definitely love robots. This episode features Founder and Chairman of iRobot, Colin Angle. He joins Alex Roy & Bryan Salesky in a conversation about America’s favorite robot, the Roomba.

Listen On
Apple
Google
Spotify
iHeart Radio
Share

View Episode Transcript

Episode Transcript

Alex Roy:

Hey everyone. This is No Parking, a podcast that cuts through the hype around self-driving cars and artificial intelligence. I’m Alex Roy, and this week, we’re going to talk about something I bought for my new apartment in Miami, which is a Roomba. A few friends made fun of me, said, “Why don’t you just clean the house yourself?” I’m like well, because Roomba exists and I’m the kind of person who buys things like that. So I bought the top of the line S9+, which is an expensive piece of kit, a little over $1,000.

Alex Roy:

And this is a really wonderful, really wonderful device. I mean, everything about it is great. You open the box, you see it. You plug the dock into the wall, you charge the robot. And the first thing it does is go out on a mapping mission of your house. And in the app, you can see it as it creates the map. And it does this so on subsequent runs when it’s cleaning, it is more efficient with time and battery, because it’s an electric device. And another great feature is you can, in the app, it’ll go into the app and block out areas where it should not go, for example a piece of furniture under which it might get stuck.

Alex Roy:

In my case, I don’t want it rolling over my speaker wires. I used to … Audio file, I got these big fat speaker wires, the Roomba would get stuck. And if the battery’s running low, the Roomba will return itself to the dock and charge, and it will also empty its own bag. All of these things are really part of a wonderful wholistic approach to product design. It didn’t happen overnight. Roomba, now the company that makes iRobot, has been around for almost 20 years. But they’re learned the lessons, they’ve listened to feedback, which brings us to my Tesla. I love my Tesla, and Tesla’s done a lot to make the electric vehicle ownership experience wholistic the way iRobot does.

Alex Roy:

But if you really get into it, it’s not as well thought out a product as the Roomba is. And it’s something as basic as the mapping function. I love my Tesla autopilot. It’s a driver assistance system, but why can’t I go into the GPS and block out areas where I do not want it to operate in case I forget to turn it off, or I wouldn’t want, say, my loved one to drive it and try to use autopilot, for example, on city streets. That would not be a good idea.

Alex Roy:

If Roomba can do it, why can’t Tesla? I wrote a column for The Drive, called Five Things My Roomba Does Better Than My Tesla. And the next thing you know, I’m exchanging emails with the iRobot corporation, which leads us to today’s guest, the founder and CEO of iRobot, Colin Angle. And joining us, because having Colin’s a big deal, of course is founder and CEO of Argo AI, my good friend, Bryan Salesky. Let’s get right into it.

Alex Roy:

Thanks so much for joining us, Colin, on the show. Really excited to have you. I was just watching an episode of Loki, this show, Loki. And Loki snatches a Roomba from a shelf and uses it as a defensive weapon.

Colin Angle:

Well, clearly Roombas are multipurpose. And being a Marvel fan, I mean, how could we not just play along when given that opportunity? But these are rugged devices. And so hopefully, it did the job for Loki. It’s tough when you’re battling against superpowers, but we felt good about that.

Alex Roy:

Is that a product placement to put it in the Loki show, or that just happened?

Colin Angle:

So we end up being part of many, many interesting things over the years, not because we necessarily seek them out, but we are sought out and we’re quite happy to show up with some Roombas and some displays and participate and just have some fun with it. But it’s not part of our funded marketing strategy, but even going back to this craziness of pets riding on Roombas, the robot has become such a phenomenon that we just get written into all sorts of shows and we say, “Okay. Well, sure.” And we’re very happy that Loki used it as a defensive weapon or device, but …

Alex Roy:

It’s kind of the mark of a very successful product though when it just sort of gets ingrained into popular culture, right? So that’s kind of cool.

Colin Angle:

I think that people love robots. And we’re supposed to have robots in our lives. And Roomba was the first robot you could actually go out to the store and buy. And in our earlier history, so much of the fact that we’re here today was because people were wanting us to win. They wanted these robots in our lives. And so there was a moment in time, the year after we launched Roomba, where we were almost completely in trouble.

Colin Angle:

We had sold about 15% of the robots that we had hoped to sell that year because we were really excited, and it was now after Black Friday, the year was almost over. We had all of this money, far more money than we could afford invested in inventory. And then suddenly, our sales went through the roof. And it wasn’t because we had done anything, it was because Pepsi put on a commercial with Dave Chappelle with a Roomba coming in …

Colin Angle:

Dave shows up in his beautiful home and then there’s a spread of chips and drinks, and he walks in and Roomba comes out and he says, “Hey, a vacuum.” And then the vacuum has this big eye on it and it spots his Pepsi and chases him around and knocks him down. And he’s on his back. His pants are pulled off, he stands up, a beautiful woman appears, and he says, “Your vacuum cleaner ate my pants.” There was nothing we could do. And we sold 300,000 Roombas in the last three weeks of 2003, and iRobot got to move on to the next year. And so, at the end of the day, having people support you, even ones that you didn’t even know would care, matter so much in the fragile world of creation and entrepreneurship.

Bryan Salesky:

No, is very true. I have to say, I always revered you and Rodney and iRobot, and especially those early years, because as someone who is just getting into robotics right around that same timeframe when you were shipping Roomba, the very first versions of it, I thought to myself, just the work that I was doing, “Will we ever make money from this?”

Bryan Salesky:

I mean, it was incredibly intellectual. I was surrounded by some of the brightest minds at CMU, and blessed to have had that experience working with them. We were doing really cool work for defense and like DARPA was kind of like always the … That money was important because they were willing to let us work on really high risk stuff. High risk meaning you just don’t know if it’ll work or not. And-

Colin Angle:

You guys built super, big, cool things.

Bryan Salesky:

… We built really big, cool things. That’s right. But there was a period of time where all of a sudden, the intellectual part of it became sort of every day. And then you started asking yourself, “But what are we actually going to do with this?” Like when can we actually build something that a customer would put it in an order for even 500, even 1,000 would have been a huge success at that point. And then I look at what you were doing, and I thought to myself, “He’s like one of the only robotics pioneers and entrepreneurs that’s actually like found a way to build a business and make money from it.” And so I followed the story very closely thinking, “Man, one day I hope to be in that position.”

Colin Angle:

Something I joke about is that part of my journey with kind of abandoning being that cool high-tech CEO, and I said, “Well, I used to be the cool, high-tech CEO.” But once I became a vacuum cleaner salesman, I started to have some success because we had to go in and torque the entire organization from being just one about pure technology, to one about saying, “Well, what do these customers really want?” And I would-

Bryan Salesky:

Well, that’s the thing. It’s easy to build a prototype and then create something that your customer can afford. And like, we did that over and over and over again. I mean, there was a project that had an incredibly lovely utility, something we’d all would want, which is an autonomous mower. And I believe it was Toro at the time. This was a long time ago. And like Toro canceled the project because like, wait a second, you mean I have to put a $5,000 IMU on this thing? Like at the time, that’s what it costs roughly. And it’s like, sorry but no consumer is going to be able to afford.

Alex Roy:

I’m sorry to interrupt, Bryan. Some of our listeners may not know what an IMU is.

Bryan Salesky:

… Inertial measurement unit, right? A sensor that can be able to tell sort of relative movements of the robot. It’s how it knew, kind of generally speaking, where it was, where it was going, what speed it was moving at. But that instrumentation at the time, this would have been, I think, the late 90s, was very expensive, very expensive. And actually, may have even been more expensive than 5,000. It was probably actually more like 20,000.

Bryan Salesky:

But the point is, it was expensive. And they were just like, even our golf course customers, they’re not going to invest in this. So I think what you guys very smartly did, was you whittled down the different applications and figured something out that met a customer need, that was actually affordable, that could actually be built. You disciplined yourself in terms of what sensors you put on the platform so that you could actually do it, actually build a real product and business around it.

Colin Angle:

I mean, to some extent we had to, because we weren’t a research institution. We were a pure play robot company. And we went six and a half years, never having enough money in the bank because we were totally unfundable. Because I mean, we were crazy robot guys.

Colin Angle:

And six and a half years, never having enough money in the bank to make payroll at the end of the month. And we just live this world of what can we possibly build that delivers more value to someone than it costs to build? And we went through like 20 different business models, things that we said, “Okay, well, this is exciting and we can get a little bit of money and it’s going nowhere.” So that the world of physical video games, where you actually battle robots, that’s a bad idea.

Colin Angle:

We got pretty far, there’s a couple ideas that we did, which were really good, but too early. Like we did some work in the energy energy business with robots that went into oil wells, so that there was a value proposition, because if you could increase the amount of oil that could be taken out of the ground, that’s huge money and that makes sense. But it didn’t work well enough.

Colin Angle:

And so that we just kept moving with is really, really harsh, but can we pay ourselves at the end of the day mindset for years. Roomba was in year 12 of iRobot, and as a result of the original Roomba, this was really fun. It ran mine hunting algorithms that we developed for DARPA, to ensure that it cleaned your home.

Colin Angle:

So we had developed these algorithms so that a robot could sweep for mines in an area. And because we thought we were going to build robots to look for mines in fields, because that’s the good thing for robots to do. But it was also a really good way to ensure that your robot cleaned every nook and cranny in your home. And the cleaning stuff we did with Rumba was derivative of work we did in industrial cleaning.

Colin Angle:

The way, Bryan, we figured out the cost problem was building robot toys with Hasbro. You got shut down on a $10,000 IMU, my equivalent knife through the it costs too much heart was actually with Hasbro where we built this really cool robot. It was called [Meep 00:14:16], and it didn’t really understand language, but if you yelled at it, it could understand the music of language, and would run away and hide. And if you said nice things to it, “Oh, good Meep,” it would come. And it was 5 cents too expensive. And they-

Alex Roy:

What year was this?

Colin Angle:

… This is like 2000.

Alex Roy:

I have to confess to both of you gentlemen, that in 2000, I was one of those people who bought the Sony AIBO robot dog. And because it said it was a fully autonomous and it could learn and I bought it, and it was not, it was not very good. It was not very autonomous. And I-

Bryan Salesky:

Did you bond with it, did it ride with you in the car?

Alex Roy:

… It had a very simple personality. If you petted its head frequently, it would make a happy iFace. And if you didn’t, make an angry face. It had a pink ball, was the only object it could recognize. And if you held it within a certain range, it would try to catch it. But if you rolled it too far away, it would just sit and the battery would die. I still have that device, but I don’t want to tell you what it cost, because I should have invested in iRobot shares instead.

Colin Angle:

But when we’re working with robot toys, the danger is to create a watch me toy. Because it’s a robot, it does these things and you can watch it and it’s minutes of fun. And then you’re like, okay I’ve watched it. The challenge is to create something that’s really more engaging. And frankly, someone is going to make a trillion dollars by nailing building a real robot pet that is emotionally responsive, isn’t hard plastic, is fun to cuddle. In my mind, that is probably the largest untapped market for robots that exist today. And somebody is going to do it.

Bryan Salesky:

I agree. A Tamagotchi doesn’t cut it for me. Like I had one, like this is not … It’s like a chia pet. It’s just not enough.

Colin Angle:

I mean, the Tamagotchi did really well.

Bryan Salesky:

I know. It did really well.

Colin Angle:

And the Furby, they sold 60 million Furbies. Half of them were to adults for keeping the adult company, not keeping the child company. I mean, there’s an unmet need there for something. And everybody names their Roomba. It’s one of these weird things that like 90% of people either use Roomba as its name, or make up some name, usually gender opposite of the name or to refer to their Roomba.

Bryan Salesky:

So in 2012, Rodney came to Google and gave a talk. And talked a little bit about the history of iRobot. And I asked him permission to take a picture of a slide that he had. I still have it on my phone and I still refer to it every now and then. Okay? He gave me permission. He said, “Collin’s 14 failed business models at iRobot,” is the title of the slide. Have you seen this? I’m sure you’ve seen this before.

Colin Angle:

It’s generous, but yes. I have absolutely seen that slide, and yup.

Bryan Salesky:

The first one is sell movie rights to and perform a robotic mission to the moon.

Colin Angle:

So that was our first business model. And we were maybe the first company to fail at private space. But we didn’t totally fail because we built a robot based on [Gangas 00:17:42], which in a weird proximity, Gangas was my undergraduate thesis in MIT, and was the inspiration for the AIBO. But we built a robot based on that, that robot. And we actually flew it out at Edwards Air Force Base in a converted Ronald Reagan era, brilliant pebble, and soft landed it.

Colin Angle:

So we put the robot in a carbon fiber cocoon, launched it, landed it in a simulated lunar environment. And this was done and funded by the ballistic missile defense organization. So the Star Wars guys, who are looking to find non-military uses for their technology. And NASA saw this and took over the project, and some of the technology wound up enabling the Sojourner rover to be added to the measure Pathfinder mission. And then as a result of the success of that mission, my name’s up on Mars on Spirit, which is fun, but it was a lousy business. So each one of these things on the 14 failed business models, they all got these crazy ass stories.

Bryan Salesky:

They’re very intellectually satisfying. Right? But the problem is the visit. Number two was something that we had looked at when we were at CMU, which was cell research robots, universities and hobbies. And of course, we did it, but there’s no margin, there’s no-

Colin Angle:

There’s no money there.

Bryan Salesky:

… There’s no money there. And number three, earn royalties on robotic toys.

Colin Angle:

Yeah.

Bryan Salesky:

Which that would have been pretty good, right? Because you can imagine all sorts of branding opportunities. Right?

Colin Angle:

But see, we did that. That was the basis, that was how we learned how to make inexpensive robots. But we inspired Alex by the Tamagotchi, where we said, “I can’t believe this is successful because it’s lame, but why don’t we make an actual baby that plays by the same rules that you have to feed and has a personality?”

Colin Angle:

And so that we made something called My Real Baby, in partnership with Hasbro. And it had the ability to make real facial expressions. It’s best facial expression was pissed off, but Hasbro didn’t let us use it because Hasbro babies never get angry, they’re just sad and happy and surprised, things like that.

Colin Angle:

We got a couple royalties, but it turns out the amount of money we spent developing My Real Baby just totally broke Hasbro’s business model for what one should spend building toys. And we think we sold a few hundred thousand of these robots, which isn’t terrible, but not good enough. And so we moved on.

Bryan Salesky:

So number 12 is develop and license a robot operating system. Well, there’s been numerous attempts at that. The problem is no roboticist wants to pay for it because they always think they can build it better themselves.

Colin Angle:

We used to say, well, Microsoft did it pretty well because the PC existed. And so there was a market for a robot operating system. I think that the big problem with robot operating systems, at least until recently, is that there’s no robots that would run them. And so we were way early on … And so that was probably just a bad idea, that had we thought about it, we would have said, “Too early.”

Bryan Salesky:

So there’s so many on the list. We’ll run it past you and we can get your permission, maybe we put the list on our website to accompany this podcast episode. So what’s interesting though, we don’t have time to go through all these, but if we were to go through all these, we would say, a) it’s super intellectually interesting. And, b) I actually think there’s a time and a place for it, but it really is a technology maturity thing and is the world ready for it thing? But none of the ideas are bad or wrong. It’s really, in my view, it’s a timing thing. What do you think, is my thesis right on that?

Colin Angle:

Most of them came out of real decent explorations. But either it was way too hard, way too early, or we missed some fundamental fact. And anytime you go and try to build a robot that replaces human labor, it’s really, really tough. That’s one of the most challenging robot industries to try to take on. So if it’s hey, I’m going to make an industrial robot to save time cleaning. Maybe you can do it, but boy, is it hard. Because those industries don’t actually want to be disrupted and making a robot that does as much as a person is usually where you trip up.

Colin Angle:

Like you can build a robot to mop floors, but if the person mopping the floor also empties the trash and goes and checks to make sure that all the freezer cases are closed and then cleans the bathroom and then leaves, well, then you really haven’t solved enough of a problem to make the robot value proposition successful. I know that because we tried it and failed for those reasons.

Bryan Salesky:

Yep. That’s right. Well, I think it was fascinating. It’s interesting to refer back to it, and I predict that there will be interesting businesses in most of those categories in the future. It’s a continuum, and it’s as technology matures, new algorithm approaches are invented, hardware costs come down and so on, right?

Bryan Salesky:

If I look at our own business in self-driving cars, we wouldn’t be here if it wasn’t for huge advancements in GPU and CPUs, huge advancements in just the amount of storage and how easy it is now to kind of collect really rich data, cloud computing that’s now available that has sort of democratized in some ways, things that were only available to a very small number of institutions years ago. So just all of that, as it becomes more ubiquitous, I think it’s going to get interesting because it does open up so many new possibilities.

Alex Roy:

There’s all these headlines, like robots are coming, robots are going to do this and that, some good, some bad. But robots have been here for, well, I guess now 20 years, you guys have been out selling Roombas. Is the Roomba the most popular consumer robot in the world, is that correct?

Colin Angle:

I think it has to be. I mean, maybe there’s some robot toys or some robot models, or there’s probably more R2-D2s out there than Roombas, but-

Alex Roy:

They can’t, right? Right?

Colin Angle:

… So it’s a question of, what’s a robot? Which is kind of a cool question to try to scratch your head around. But yeah. I mean, Roomba is the first practical robot, and well over 30 million robots are out there and-

Alex Roy:

30 million Roombas.

Colin Angle:

… 30 million Roombas. And the crazy-

Alex Roy:

Wow.

Colin Angle:

… thing is for the last … I don’t know, four or five years, Roomba has been the number one selling vacuum cleaner in the US, which to me was like, what?

Alex Roy:

More than Dyson, like really? Really? Really?

Colin Angle:

It’s the number one selling vacuum cleaner in the country. And that’s pretty crazy. And so that I know going back to our earlier conversation, it’s like, when I imagined when I founded iRobot back in 1990, it’s like, if you said, “Okay, it’s going to be 31 years later. And 90% of your revenue is coming from a vacuum cleaner.” I would like be completely bummed that it’s taken so long to get to where we are today, but also blown away that suddenly, we’ve obsoleted the traditional upright vacuum cleaner, and we sold 30 million robots. So the world is definitely different and stranger than I would’ve imagined.

Alex Roy:

I mean, were there attempts before iRobot to build devices like that? Because the Jetsons had all manner of robots cleaning up in the 60s.

Colin Angle:

Yeah. I think the earliest robot vacuum cleaner, I saw a picture, world’s fair. I think it was in 1960-something. It wasn’t round, it was kind of trapezoidal shaped device. I think it was one of the radio companies that had made it as part of the house of the future. And of course, the Jetsons, since 1962, they had Rosie, the robot pushing an upright vacuum cleaner, which really confused people for decades because people thought that that’s what robot vacuuming was, was to build Androids that pushed traditional vacuums.

Colin Angle:

And when we launched Roomba, we didn’t even call it a robot because people didn’t think it was a robot because robots were Androids. And so you just made an automatic vacuum. It was actually the press that called it a robot.

Alex Roy:

It’s funny because there’s-

Bryan Salesky:

I didn’t know that. That’s really fascinating. Yeah. It is interesting how when you use the term robot, everybody has a slightly different vision for what they expect that to be. But I think at this point, it’s pretty undisputed that Roomba is a robot. And people have taken them in almost like pets in their house. I mean, it’s amazing how friends and family that have them, it’s like part of the family. And if it’s not functioning properly, it’s like a major problem because it’s part of their lifestyle now and the kids love it and the pets love it. And it’s sort of part of the routine.

Colin Angle:

This was one of the first completely bizarre things that happened to us after we launched Roomba. And it still happens today. In the early days, we all shared customer service and it’s because I wanted all the engineers to know what was going on. I wanted to know what’s going on, so we all manned the phones.

Bryan Salesky:

That’s awesome.

Colin Angle:

And so I remember this really vividly. I get this customer complaint call, and I said, “Okay. Yes, ma’am. Okay. Your Roomba doesn’t work. Okay. Yes. And yeah, it’s the wheel module. I get it. Sorry. We’ll send you a new one.” And she says, “No. No, I’m not sending you Rosie.” She wanted the white ambulance truck with a guy with the lab coat to come in and perform open-heart Roomba surgery on Rosie-

Colin Angle:

… because she had this connection to Rosie. And if she sent Rosie back, well, it’s like killing Rosie. I mean, I don’t want a new Rosie. It’s this connection that develops is really bizarre.

Bryan Salesky:

I think it’s really cool though. And so that shows you the power of like really well done product design, right?

Alex Roy:

I got to tell you, I have a Roomba and a Braava, your mopping product. And my two-year-old, when she wants to be the one to activate them, and then she wants to follow them around. And if they do something different on a certain day, she’s very excited about it. And then they go home and she’s like, “Oh, they’re home.” And she’s very, very connected to them. It’s a real big product design win.

Bryan Salesky:

Can you imagine calling product support and getting Collin Angle answering? That’s awesome. But I have the same instincts though, because when we launch our first product, you’ll be sure that I am going to be there near the phones to see what’s happening in the car when someone presses the, I need help button. I’ll for sure be there and I will definitely take one of those calls all day long.

Colin Angle:

And when things go wrong, it is like the biggest sucker punch in the gut. You’re like, that’s not what’s supposed to happen. But it is motivating to try to make it right.

Alex Roy:

So here’s a question.

Colin Angle:

All right.

Alex Roy:

When today, I have a Roomba, I think it’s an S9+, which is the top of the line, like it’s the best of the best, state-of-the-art. And it goes back to its base and it can empty itself … It empties its trash into this bag. It is as seamless and holistically perfect a product as I’d want it to be. But if it didn’t have the bag emptying, and I had to go empty the bag, I would pull my hair out. But when you launched, you didn’t have the bag emptying feature. Correct?

Colin Angle:

Mm-hmm (affirmative).

Alex Roy:

All right. So what was the minimum viable product? What was the product that hit it out of the ballpark? Because Chappelle does a video, and suddenly you sell 300,000 of them. What did that thing do on day one?

Colin Angle:

So we had this algorithm, the robot didn’t know where it was back then, but it had strategies to sort of cause it to get places. So it would clean an area and then it would kind of randomize its location and then do another crop circle. And then also had the ability to go and clean along the edge. So we didn’t know how to localize our precise position back then, not at the cost that we knew we had to hit. But we had created this heuristic strategy for getting everywhere. And so then we needed to say, “Okay, well clearly people don’t want to vacuum their entire house every time they turn it on, so we need to constrain where it cleans.”

Colin Angle:

So we created this … This might’ve been one of the smartest things that we did, the virtual wall. So there was a little, it looked like a coffee cup, and it shot out an infrared beam that the robot wouldn’t cross. And so that was the next thing. And then we need to figure out how to make it clean well enough. And Roomba actually started out as a robot that would drag an electric static cloth, like a Swiffer cloth. But then when we talked to customers, an automatic Swiffer robot, they were only willing to pay 50 bucks for, but if it was a vacuum, my boy, they would spend way more money on a vacuum than a …

Colin Angle:

So he’s like, oh crap, we’ve got three months to add a vacuum to our robot that was supposed to only be a sweeper before. And so we had to invent the world’s most efficient vacuum. So it’s kind of like when you have a hose and you stick your finger in front of the hose so it squirts faster so you can wreak havoc on your siblings and other such things or maybe wash your car? We ended up taking a really wimpy vacuum cleaner, and then running that vacuum through two narrow sheets of rubber to accelerate the air, to get enough velocity air so that the air would hit the small particulate and suck it up.

Colin Angle:

Of course, that didn’t work at all for big particulate, but you had the sweeper in front of the vacuum. And so that we had accidentally, by desperation and need of innovation, decomposed vacuuming into a two-stage process, where you picked up with big brushes, the popcorn and the big stuff. And then the fine dirt was this innovative squeegee vac. And we could legitimately call it a vacuum cleaner. And we had this thing that went around your home, was blocked by the virtual wall, did a really good job vacuuming and surprised everybody that brought it home who was certain that this was just a silly anecdotal gadget. And then wow, it kind of actually works and that’s awesome. And that was the beginning.

Bryan Salesky:

You must’ve had a really good robot proving ground in your lab with like every bit of type of debris and what.

Colin Angle:

I mean, we had the magic recipe of dirt. But man, we screwed up so many times in the early days, just to be clear. Because we designed this robot to what we thought was like the awesome gold standard of quality, which was the European uprights will operate 150 hours without problems. So it was okay, we’ll just do that. And then completely miss the fact that, well, that works great if you vacuum half an hour a week, that lasts a long time, but if you’re vacuuming an hour every day, that’s about five months. And so that we had the entire … People were on their fifth Roomba before we fixed a subsequent generation, and we just said, “Okay, this is great. We’ll invest in customer service, and if your Roomba breaks, we’ll just send you a new one and keep you loyal.”

Colin Angle:

And there’s this really cool thing that is the savior of many small companies trying to figure it out, which is you have more customer loyalty if one of your customers has a problem and you treat them well and solve it for them, than if they never have a problem, because they actually know that you’re standing behind your product. And that one, that’s another on the list of reasons why iRobot exists today, is that we had tons of problems, but we stood behind the people and our customers, treated them well, gave a new product and bought ourselves enough time to say, “Oh, why didn’t we think of that? We’d better go and have a crash project to fix that,” and then iterate and iterate.

Bryan Salesky:

What is the oldest Roomba that is in operation with a customer today?

Colin Angle:

There’s definitely people who still have working original Roombas. And if you don’t beat them to death and if you upgraded but then realized maybe I’ll hold onto this robot as a bit of history, there’s no reason why your Roomba can’t still work. You might need a new battery, or if you didn’t leave your … It was a Ni-Cd battery at the time, and deep discharge eats those things. So if you kept it charged, you could have an original Roomba that still worked, and there are a lot out there.

Bryan Salesky:

Well, going back to what you were saying earlier on customer service, I think it’s a big lesson for any company getting started out, and I’m glad you shared it. I was telling our company, some people yesterday that it’s one thing to say that … Like a lot of companies will say that they have a customer obsession, they want to obsess about the customer and their experience, but it’s one thing to say it, but it’s another thing to actually do it. And I think it’s that type of mindset and that execution and detail oriented-ness towards the customer that really makes the difference between sort of a mediocre product and one that’s incredible. And even beyond that, as you were saying, a loyal customer, earning that loyalty, I think is an important lesson. Can you share with us what you think the future of household robotics is, Colin?

Colin Angle:

Let’s face it, the smart home today isn’t smart. It’s a chaotic mess of stuff.

Bryan Salesky:

There’s no standardization, it’s terrible. Like just-

Colin Angle:

Well, I mean, we’ve got like 70 different apps. Maybe I’m not representative, but I’ve got more apps than I can imagine on my phone. The house is full of stuff. And if we think back to the days of the early VCR and the fact that 97.5% of those VCRs had the blinking 12 clock, because no one bothered to program the clock. Well, humanity, hasn’t gotten smarter in their ability to program. And so that this idea that there’s this huge burden on the end user to figure out how to program up their home to make it smart appeals to like 12 people. And the rest of us are stuck scratching our heads. And we bought this really cool gadget, it works. We researched, we got it up and running and then if we ever had to touch it again, we’re going to be really mad because then we have to relearn everything having to do with how to set it up and make it work.

Colin Angle:

And so, we have a problem. What we need is to think about the smart home more like we think about a robot, than we think about a place to put stuff. We need to think about it as a system. And Bryan, you might think this is cool. Back in the day, robots needed to be self-contained. Then with wireless connectivity and the cloud, it’s like well, part of the robot’s brain could sit up in the cloud and that would give us access to do a lot more. And so that maybe the brain could be outside the robot. But now with ubiquitous Bluetooth LE and low latency wireless connectivity of all sorts, well why does the robot even need to have a body? Why can’t I have my eye on my front door, my muscle in the door lock, a muscle to raise and lower the shades? Say, skin sensor is my thermostat in each room, the lighting systems. And why not just explode the body of the robot, but still think about the overall system being aware of its own configuration? So it knows where all of its muscles and eyes and ears are, and body, it knows what rooms are and so forth.

Colin Angle:

And with enough of that context and knowledge, learn what it’s supposed to do on its own so you don’t have to program it. So instead of programming, maybe you have preferences. Maybe there’s a set of rules that your home is supposed to follow. And every device that gets plugged in instantly inherits the rules that are set up for that home.

Bryan Salesky:

I mean, when you think about it, everybody has their routine. And it varies if you’re in the weekend or if you’re coming home from work at night or if you’re having visitors over. And even if there was a way to just sort of like have the whole system work in concert through those different regimes, like it doesn’t even necessarily … This is the thing. Because like look, I want to apply AI like just like the next guy and have it learn all this fancy stuff. But at the end of the day, like I’m fairly, and I think a lot of people are, fairly predictable, depending on what the scenario or the context is. So even just being able to link together the lights, the shades, the thermostat and all of that into a smart system that then is context aware of whatever situation we’re in, that’d be huge. Right?

Colin Angle:

Right. And the fact that we are creatures of habit means you don’t need breakthroughs in AI to do this. You need breakthroughs in system engineering or robot engineering to go make it all just program itself and follow a set of preferences that are reasonably easy to articulate, and make the smart home a self-configuring bit of thoughtful system tech.

Alex Roy:

So I have top of line Roomba S9+, love it. I have a Braava, the robot mop. Love it. Now, you talk about this feedback loop, you really listen to your customers.

Colin Angle:

Yeah.

Alex Roy:

Okay, so-

Colin Angle:

They can tell me, they can share their maps.

Alex Roy:

… Well, yes.

Colin Angle:

Okay.

Alex Roy:

Let’s be really clear here. Okay? One of the things I get to do at Argo AI is I get to go to Miami beach, I ride around in the autonomous test vehicles, and I have an app. And in that app, I can enter feedback. Oh, I think that I really like this, but it could do this. Can it go here? And then within a few days or weeks, it does.

Colin Angle:

Yup.

Alex Roy:

I submitted a ticket precisely on the issue you suggested. Why? I have this lovely map. I almost wish you could see it on my phone. I have my vacuum map and I have my mop map. I submitted a ticket months ago. When will the … And your app is great. I mean, I’ll-

Bryan Salesky:

Are you using this as an opportunity to get priority on your product request for Colin?

Alex Roy:

I see that people in your team have already integrated some components of the functionality of the Roomba, the Braava, because I can program them to vacuum first and mop after.

Colin Angle:

Yeah.

Alex Roy:

Why aren’t the maps integrated yet? My ticket was in months ago.

Colin Angle:

Of course, we need to do that. And we can certainly give you a tagline and credit on it, but shared maps is something that I’ve wanted maybe for more than just a couple of months. So I’d probably predate your ticket by maybe about five years, at least as far as investment of actual resources.

Alex Roy:

Five years.

Colin Angle:

So I mean, yes. I mean, you asked me what is the future of the smart home. So I got to like turn my future-O-matic machine forward. But we need the technology to be in service of the customer, we need make sure that the understanding of the home is shared across multiple devices and not just Roombas, but other things in your home that care about what room they’re in. They should be able to benefit from that information too. I’m actually psyched I got your ticket right, because I didn’t know that, but is one of CEO, Colin Angle, of iRobot’s major peeves, that we don’t share maps between our devices yet, because that’s part of this seamless future that we need to aspire to get to.

Colin Angle:

Because if the home is a robot, everything needs to know where it is so that we can say, “Okay, we got the lights in these rooms, we got these in these rooms. We got these shades, these thermostats.” Maybe the Roomba is the white blood cell of the home as a robot. And it’s going around making sure everything is as it’s supposed to be. And we can do it. I mean, this is not gee, one day we will have jet packs. This is more like this is a thoughtful, smart home that we have the technology today, if we have the will to go put in the work to make it all work together. And it’s going to give us a home that maintains itself, operates more efficiently, gives you a secure environment. And what I’m passionate about is how do we make homes healthier places to live through the use of technology as my foot stomping future goal for where we’re headed?

Bryan Salesky:

I think everybody could buy into something like that. And I wish you a ton of luck in pulling all of that together. The home automation environment today is a mess as of my own research. I tried to get into it myself a little bit. And I think I have five different protocols running at home with a Linux server with a bunch of custom stuff that tries to like tie it all together. And that is not what most homeowners are going to be able to deploy. Right? Like there’s a huge standards, like an interoperability issue that I think it would be cool if iRobot could lead in that conversation and like pull these things together so that there’s sort of a common operating system for the home.

Colin Angle:

It needs to bring ease of use and some amount of real understanding so that what is being asked of the occupant of the home is not what I have in my home, right behind the wall behind there, I’ve got a server rack to try to go and operate my smart home, which is a ridiculous thing to have to have in your home. And if you want it to be-

Bryan Salesky:

It sounds like I need to get some tips from you, Colin.

Colin Angle:

… No-

Alex Roy:

I’m going to connect to you guys.

Colin Angle:

… Yeah-

Bryan Salesky:

Z-Wave lights, Zigbee curtains, or like this thing that like is able to automate your roller shades or whatever, like OpenHAB to like help pull those two things together. And then some sort of extension that then can publish all those devices through whatever it’s called, the Apple home automation home kit, so that you can then use the Apple app to be able to turn all that stuff on and off.

Bryan Salesky:

I successfully was able, Colin, to pull all of those things together. Not to mention the other connector that then goes from that server to the cloud, because it has to be in the cloud for it then to talk to Alexa so that Alexa can then voice activated turn my lights on and off. Yes, I managed to pull that off.

Colin Angle:

So Bryan is one of the 12 that know how to make the smart home work today, but …

Bryan Salesky:

If that’s what it takes, iRobot, please solve it.

Alex Roy:

Look, the average person is closer to me than to you gentlemen, literally, robotics guys and rocket scientists. And a lot of people ask me, “Like when,” just like they ask, “When is self-driving cars going to arrive?” And I say, “Well, they’re here, and not fair yet, they’re coming.” When is a Roomba going to be able to clean up, say, a little bit of dog poop? It’s a basic question. Like it’s a basic use case.

Bryan Salesky:

Alex, or … I’m just curious.

Alex Roy:

No. The average person doesn’t think about technology. They think about problems they want solved, and bite-size chunks. When will it do this or that? When will a Roomba or a Braava, in combination, be able to deal with a pet.

Colin Angle:

So step one is not cleaning the dog poop. And then step two would be okay, well, what is the right strategy for cleaning dog poop? I mean, because it might be that I need a manipulator. But happily, machine vision is getting to the point where we can do all sorts of really cool stuff, because for the longest time, iRobot had really cool, low-cost manipulation technology, and nothing to do with it because we didn’t know anything was. And so what’s the point of being able to say, “Go to the kitchen and get me a beer,” if you don’t know where the kitchen is? Much less the refrigerator. And so that we’re right on the cusp of knowing enough stuff with recent advances in lower-cost machine vision for-

Alex Roy:

For the Roomba not to run over dog poop.

Colin Angle:

… Well. I mean, there’s that, and then there’s just knowing where stuff is. And then once you know where stuff is, then you get into this next really cool problem is well, what do I do once I know where objects are? And Roomba and dog poop, sure. Step one, avoid it. Step two, send out the emergency response robot to go deal with said hazards, if you want to actually pay for that level of dexterity in your home, or maybe you have to wait for a more general purpose robot in your home that has the capabilities of picking things up that need to be picked up.

Colin Angle:

But that’s a hard one, but I do think that your robots should move from just doing the same thing everywhere, to taking a different approach to whatever it is they’re supposed to do based on the environment that we’re in. And again, the slow moving, but accelerating engine of home awareness is starting to give us enough knowledge to take on those types of tasks.

Alex Roy:

Have you ridden in an autonomous vehicle, and what do you think?

Colin Angle:

I have. Not for a long time because it hasn’t been available, that I’ve had access to. But I rode in one of the early Google autonomous vehicles back at an event 10 years ago, and we were on like a gravel track and they had it doing rally car racing. So it was sort of meant to scare the crap out of you and make you, do you believe that the technology is going to work or not? And it was awesome. So I saw a huge potential then.

Colin Angle:

What do I think now? I think that the devil is always in the details and that we’re going to see a steady progression of increasing safe utility of autonomous vehicles where the vehicles know enough and the AI gets smart enough to start passing the safety thresholds that we need. And gosh, the data is being collected at such a huge rate at this point, that I think we probably have surpassed human-level driving safety or close. I don’t know, Bryan, you can correct me on that.

Bryan Salesky:

I think we have in certain conditions and certain types of … Like what do I want to say? Certain scenarios, right? Like across the board, like what an average human is able to do across all seasons, all weather, all different types of streets and roads and so on, no. But absolutely when it comes to like really challenging city driving, where it can become really overwhelming to even an average driver, like an average good driver’s senses, I think that we have surpassed those capabilities.

Bryan Salesky:

Simple fact that we don’t get distracted, we’re always learning. We’re able to see so many more things, and reason about what each and every individual actor is doing many times a second is a … Humans are not able to do that.

Colin Angle:

At the end of the day, I think that autonomous driving is inevitable, to be a complete viable replacement. And I think it’s exciting that it means that we can drive for fun.

Bryan Salesky:

And do other things while you’re getting there. Most people are bad at putting in a value on their time, but everyone will say regardless of what value they put on, everyone will say they want more time back in their day, right? And so if you want to drive and you enjoy an adventure then drive, we’re all about that freedom of choice. But most people want to do that from a more recreational perspective. Their every day commute, or night on the town, whatever it is, they don’t really want to drive.

Alex Roy:

What lesson can autonomous vehicle developers learn from the history of iRobot?

Colin Angle:

Everything takes longer than you want it to. I think we touched on it a little bit previously around the problem may not be exactly what you think it is. Meaning that solving autonomous navigation actually creates secondary opportunities to make the experience of autonomous driving deliver on the value that people imagine it will. So the idea of that why do I want to autonomously drive? And it’s so that I can do other things. So then the question is, well, are you spending as much time on enabling the doing of the things as you are on the autonomous driving? And that I think companies that think through the whole problem are going to be very differentiated from the isolated area that where attention is today. And one of the things that I learned recently, a few years ago, where I went on stage and had been going on stage for many years and people say, “Well, Colin, what is the ultimate Roomba?”

Colin Angle:

And I would say, “Well, the ultimate Roomba is the Roomba that you never touch, you never see. You just come home every day to a perfectly vacuumed home.” And boy, was I wrong. Because what people actually want is control. And they explained it to me in this completely brutal way, where they said, “Well, imagine you had a cleaning person in your home that you couldn’t talk to, and you just had to trust that they would go and clean your house the way that you wanted it, when you wanted it clean, in the manner. And that person probably would last about 15 minutes before they were fired, because we all want our house clean the way we want it, at least to a certain degree. And the inability to have that level of control is maddening. And so that if you want to make a great home robot, you need to think about not just autonomy, but about collaborative autonomy.”

Colin Angle:

How do you work in partnership with the owner of the robot? And I think, I haven’t heard a lot of talk about, in the autonomous driving vehicle space, about that question. This is a really hard problem, autonomous driving and people are really focused on how do I make it completely autonomous so I can get in and it can drive me to my destination?

Colin Angle:

But again, if you’re being driven somewhere and you can’t talk to the person driving the car, is that a great experience? And I think that there’s some lessons there that as the industry matures beyond minimum viable don’t crash, into how do we make autonomous driving the experience? Something that plays an incredibly valuable and important role in our lives, there’s a lot more that will need to be done.

Bryan Salesky:

Colin, thank you for doing this. I think we’ve covered some really interesting territory, and you have an open invite to come and enjoy an Argo self-driving car anytime you want to experience it, so …

Colin Angle:

I would like to do that. Sure.

Alex Roy:

Well, that was a wonderful episode. What really struck me was … Well, first of all, I didn’t know the history of iRobot. I mean, they were founded in the 90s, and how close they came to not making it. And if it weren’t for that David Chappelle ad with Pepsi, they might not have made it.

Alex Roy:

And there’s this notion of inevitability around technology, and we’re surrounded by devices we take for granted. And I would argue that technological progress is inevitable, but the companies and brands that survive that sell them to us, they are not. And if you go back to the history of the car or elevators, electricity, you’ll find decades, if not centuries, of consolidation around modes of power transmission, and transportation, and a lot of companies didn’t make it. The ones we know of, the ones we use every day, the brands we admire, those were the last ones standing and the true survivors.

Alex Roy:

Any case, if you enjoyed today’s episode, please connect with us on social. We’re on Twitter @noparkingpod. I’m everywhere on social @alexroy144, but especially on Twitter @alexroy144. It’s the numbers, 144. Please share No Parking with a friend. Like us, subscribe, give us five star reviews wherever you listen to your podcasts. This show is managed by The Civic Entertainment Group. And until next time, I’m Alex Roy, and this is The No Parking Podcast.