BotJunkie Interview: Evolution Robotics’ Paolo Pirjanian

Writing by Evan Ackerman on Friday, 15 of October , 2010 at 1:14 am

As we commented in our review, Mint takes a different approach to robotic cleaning than we’ve seen before: it eschews the vacuum for simple, silent, and efficient sweeping and mopping. Despite its simplicity, however, Mint is also able to localize itself, which is a pretty neat trick for a $250 robot. With all this in mind, we had some questions about how Mint came to be as well as what the future holds, and Evolution Robotics’ President and CEO Dr. Paolo Pirjanian was kind enough to help satisfy our curiosity. It’s a long interview packed with interesting info on the development process for Mint and the NorthStar system (as well as hints about what’s next), and you can read it all, after the jump.

Tell us about the history of Mint… What inspired you to create it, and what made you decide that now was a good time to introduce a new type of cleaning robot into the market?

It was a combination of things. There was the maturity of our NorthStar technology combined with our additional software system that allows us to have a very reliable coverage system, and we saw that there was a need in the marketplace for cleaning hard surface floors particularly. In the robotic vacuum domain, Roomba and many other entrants are already covering that space, so we didn’t feel that there was a need to go and make the [vacuuming] space even more crowded. Also, the Swiffer type of floor cleaners have become really popular… They represent a billion dollar market. The core value of the Swiffer type cleaner is about effortless cleaning, where you don’t have to bring your broom or your mop out but you just have this cloth that you can use and throw away, or you have microfiber cloths which are reusable. Mint furthers that whole idea of effortless cleaning where you basically press run, and you’re done, so it takes the chore of floor cleaning out so that you can use your time for something you enjoy doing, rather than pushing a broom around.

The other, more personal history behind this is that I moved into a new house about three years ago, and we have really dark hardwood floors throughout the whole house and white tile in the kitchen. We realized that keeping the floors clean is a constant battle… I mean, we could have spent an hour cleaning the floors and five minutes later my son and his friends would come from soccer practice and run through the house dragging the dirt and grass in, and then you have to go back and get the Swiffer or the vacuum cleaner out and do it again. My wife and I spent a lot of time fighting to keep those floors clean, and because we have these really dark hardwood floors, everything shows on it. And in order to keep the floors clean, we had to use an artillery of different products. My wife had obviously seen me, throughout the last ten years or so, bringing various types of robotic cleaners home, but she never adopted any of those products. They were loud, they weren’t really doing their job, and after they cleaned you had to bring out your manual cleaners anyway. So, all of this combined in my mind in terms of, it would be great to have a Swiffer type product that really cleans hardwood floors.

One important thing that I noticed with my wife is that with these other products, when you turn them on you really have to leave the room because they’re loud. So we emphasized that we wanted this to be something that is not intrusive to your home. You can speak on the phone, you can watch TV, you can do homework with your kids: it had to be quiet. Other things we noticed when people are cleaning is that these round shaped robots can’t really get into corners which is where, especially with hardwood floors, the air circulation pushes all the dust into the corners and underneath furniture and so on and these round robots are not really able to get there. So, all of these things we’d learned over the years led to the idea for a square robot that’s silent, very easy to use, no maintenance required, because we’d also noticed that with some robots and manual cleaners you have to keep cleaning the brush and so on. So anyway, this is basically a flashback of the last ten years of this experience that culminated in the concept for Mint.

After that, we spent a lot of time in focus group studies. We brought in people who we considered to be the target market for Mint, which is women from 25-54 years old, described the concept to them, tried out various things, and finally settled on the set of features here that we had to deliver on. We spent the last 18 months working on refining Mint to get it to the marketplace.

Mint is very simple compared to other cleaning robots. What made you decide on this essential set of features?

Mint has a set of features that we considered very carefully, and we did a lot of market research (both quantitative and qualitative) to try to understand what the features are that people really care about. For instance, Mint has the ability to localize itself so it knows where it is all the time, so getting back to some kind of charging station from a technical perspective would be a no-brainer to add in there. But, we learned that people really didn’t feel that that was an important feature, especially if it adds cost to the product, which obviously it does. A lot of people wanted to just treat Mint like a regular floor cleaner, and once it was done with its job, tuck it away and forget about it, rather than have it sitting in the middle of the living room or the kitchen where it occupies some space. The other aspect of it is that when we talked to people who had experience with robot floor cleaners that had self docking, they said you have to clean the vacuum canister, you have to clean the brushes, so it’s not like it’s fully autonomous. And in many cases, these robots do get stuck… With Mint, we tried hard to minimize the possibilities for it getting stuck, but it will, eventually, in some specific scenarios, get stuck. So the promise of the fully autonomous robot that you can just set up and forget about is not here yet. From people’s perspective, [the docking] was viewed mostly as a gimmicky function, which they didn’t find much benefit in. We decided that rather than focus on things that are gimmicky, we’d spend our time and effort focusing on refining certain things like edge cleaning and getting into tight spaces.

So, it was an informed decision about the trade-offs that we learned by talking to the consumers about, and understanding what features they really see benefit in. Now, from a marketing perspective, I can see us being compared head to head, feature by feature with other products where people will say, “Mint doesn’t have self-docking or self-scheduling” and so on. And I think yes, on marketing we may get a little bit of negative points on that. But I think once the users get it into their homes and use it they will completely get the product, and they will I think appreciate the design trade-offs that we have made. And that’s really the background for why we went that direction.

How has Mint been received as a commercial product so far?

Mint has just been launched on the Home Shopping Network and is now available at Bed Bath & Beyond. I feel like those are very positive signs for a company that hasn’t been in the consumer product space before. Being able to get partners like that is a big win for us… I think the reason for that is that these retailers have seen the potential for a product like this, and it seems that the consumers who are using the product are satisfied with the product too. On Amazon we already have people writing reviews on it, and the reviews seem to be very positive. People understand the product, and they find it useful. This is our first quarter for the launch of Mint, and so far one of the biggest challenges we have had to deal with is supply pressures. We take all of these things as positive signs for the prospects of Mint, but the jury is out of course… We’ll have to live through this quarter and see how demand keeps building up for the coming years. So, we can’t declare victory as of yet, but all of the indicators are trending in the right direction, which we are very happy to see.

One of the things unique to Evolution Robotics and Mint is the NorthStar localization system. Can you talk about how you came to develop that system, and what the future for home robotics localization might have in store?

When we started Evolution Robotics years ago, we started working on localization as one of the key enabling technologies, because for a robot to be smart, and do interesting tasks in its environment, it really needs to be able to navigate from point A to point B. The first challenge in that is for the robot to know where it is, because if you don’t know where you are, you can’t really figure out how to go where you want to go. We see localization as a core building block. Mint shows how this building block can be integrated seamlessly for the user… Basically, you put the cube up and you forget about it.

When we started off in 2001, localization technologies were just beginning to be available in academia. At Carnegie Mellon University, a couple professors including Sebastian Thrun and Dieter Fox were the first to come up with a solution for localization, which used a ten thousand dollar sensor and a bunch of computers to build a map of the environment and do navigation. Gradually, those technologies have been making their way into industrial automation for palletizers and those kinds of products. Evolution Robots started by saying, we want to come up with solutions for localization that are cheap enough for us to put in consumer products.

So, we saw this as an opportunity, and spent years perfecting the technology and reducing the cost to a point where we went from a solution that would cost tens of thousands of dollars to something that would cost on the order of simply dollars so that we could put it into a consumer product. That path has taken a lot of tenacity and persistence to get there, because all the odds were stacked against us and it looked like mission impossible… I even recall a few years ago, I have a scientific advisory board of prominent professors in robotics from around the world, and their message to me was basically, “just give up, you can’t do this.” But, we did. And we were very proud about doing it when we introduced it in Rovio for the first time. But, we were really not satisfied with the solution in Rovio for various reasons. For example, you have to plug the Rovio beacons into outlets, and so with Mint we tried to perfect the user experience. You can just place Mint’s beacon anywhere in the environment, and the robot will ping it to turn it on and off for you. It’s non-intrusive, and the pinging function allows you to save battery power on the beacon, because we wanted the battery to last for months.

For Mint, in the future you can imagine having multiple cubes so that the robot can clean beyond the current limitation of about 1000 square feet… You could have a couple cubes to allow the robot to navigate throughout the whole house automatically. Beyond Mint, you could imagine other types of robots in the home or business that need to be able to navigate in the environment intelligently, and then NorthStar and NorthStar type technologies become a very important building block. Far in the future, we see NorthStar or future generations of NorthStar which may not require even a beacon in the environment, and may just use what is available already in the home to navigate. In addition to NorthStar, we have another technology which is known as V-SLAM. It uses a camera to visually map the environment using pictures, and then uses that map to localize the robot very accurately. These kinds of technologies become building blocks that may open up opportunities for building other types of robots than just floor cleaners.

On iRobot’s ‘celestial navigation’ patent:

I guess when a company develops a new technology it does inspire other companies to try to do something similar. One of the things that you may be able to look at is that Evolution Robotics since 2001 has filed a number of patents in the localization domain including NorthStar, which predates all of these other technologies. On one hand, I think it’s flattering to see other companies such as iRobot, who is currently viewed as the biggest consumer robotics company, looking at us and at some of our technologies and trying to do something similar. On the other hand, I can say it’s one thing to come up with the idea, it’s a completely different thing to get it to work. And specifically, getting it to work within the particular size or cost targets that we need it to be at.

We’ve had some questions from people who are potentially interested in using the technologies inside Mint for their own projects. Would you consider offering any components, such as the NorthStar system, as modular accessories?

For Mint particularly, we tried to focus on an appliance grade of product. You’ve noticed that there’s a USB port, which is there primarily for firmware upgrades and so on. But, the USB port could also give access to the hacker community to start trying to, if you will, reverse engineer the APIs. We currently don’t have any plans to provide a developer kit for Mint, but if the hacker community shows enough interest, then we would definitely be there to support them, and maybe to provide them with more insight into the internals of Mint to enable them to do what they want to do. This is primarily from the perspective of encouraging the developer community to start tinkering with robotics applications, which I think overall helps the robotics market. In terms or providing NorthStar as a module in the future, that’s possible too… I guess we would wait to see if there’s enough interest in the developer community, and if that’s the case it would be in our interests to enable them and to allow them to be successful in what they’re trying to do.

What does the future hold for Evolution Robotics and Mint?

Regarding Mint, our plan is to focus on hard surface floor cleaning, and Mint is our first entry into that market. Assuming that it’s a success, we will continue to introduce next generations of the product with functions to provide better cleaning. Our ultimate vision, or the way I describe it in a very simple sentence, is “no more dirty floors.” This means that our vision for robotic floor care is that within the next ten to fifteen years I think none of us will be spending any of our time cleaning floors: we will have automatic floor cleaners in our home that will continuously keep our floors clean, both hardwood and carpet. It would be like a dishwasher, if you will… You install it once in your home and you forget about it and it basically just keeps your floors clean.

From a technical perspective, in terms of robotics technology, I think we are there… We have a very capable navigation technology. The improvements that have to come relate to cleaning mechanisms that can do a great job of cleaning. Mint is using Swiffer technology, which is for maintenance purposes primarily, and if you want to be able to clean mud and thick stains there are still improvements that we can make. The other portion of it that requires improvement is related to a fully autonomous system where you can have a docking station that can replenish the robot both in terms of power and emptying any dirty fluids or cloths or dirt, allowing the robot to go on on its own. You can imagine that this is probably a device or appliance that’s connected to power and water and sewer, like your dishwasher, while the robot is like a printer head moving back and forth on the surface of your floors, constantly maintaining them. So, that’s our vision for it, and by launching Mint and listening to the feedback we’re receiving from end users, I’m sure we will learn a lot in terms of what capabilities they are looking for, and we’ll include those things in a future generation of our product.

Many thanks to Evolution Robotics and Dr. Paolo Pirjanian for speaking with us.

[ Evolution Robotics ]
[ Mint ]

Comments (4)

Category: Consumer

• Add this post to
• Del.icio.us -
• Digg

Name (required)

Mail (will not be published) (required)

Website

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>