7 min read

Attack of the robot floor cleaners

It seems that in the near future we can expect any surface to become a touch screen

My friend Anne-Marie Merritt has an amusing tale about her Mac getting pregnant in her newly created blog “Context Sensitive” (with the wonderful URL “argc.blogspot.com”), and I thought the posting may amuse you Mac people.

So back to serious stuff: It seems that in the near future we can expect any surface to become a touch screen. And not just walls and glass, but even unexpected surfaces such as ice.

Yes, Nokia’s Context-Aware Social Media team (now there’s a cool-sounding job) at the company’s research lab in Tampere, Norway, built a touchscreen out of blocks of ice. They called it Ubice, or ubiquitous ice (I’m not sure what that actually means, but whatever).

The Ubice system was built of blocks of river ice assembled into a 2 meter by 1.5 meter wall and the location of users’ hands were detected using rear-diffused illumination (RDI), which was the basis for Microsoft ‘s (NASDAQ: MSFT) Surface, a touchscreen table that you can purchase for the princely sum of $12,500.

RDI works by projecting a plane of near-infrared light focused on the touch surface; in this case, from behind the ice wall onto the user side. When an object (a hand or anything else opaque) intersects with the illuminated surface, cameras operating in the same part of the spectrum detect the object’s location, size, and movement whereupon regular, visible spectrum projectors paint images that are also focused on the user side of the ice wall.

Some of the imagery Nokia used for Ubice were flames and fire creating, dare I suggest, a poetic image of contrasting primality … or something like that. Anyway, it’s dead sexy but somehow I can’t see this being used in any enterprise that isn’t in something like the cold storage business.

Now, shifting gears dramatically, I’ve always been intrigued by robots and, while there are many amazingly humanoid designs that can do amazingly clever stuff (for amazingly high prices), the only robots commonly available today are of the household floor cleaning variety, so I figured it was time to take a look at how far our future overlords have come.

Back in the second quarter of this year I moved into a house and the new abode has a main “great” room with a wooden floor. This floor is very smooth and requires constant mopping (I’ve never had a floor that shows up dust and dirt so easily) and constant vacuuming (we have two large golden retrievers that shed enough hair each week to stuff a decent sized sofa).

To make this onerous maintenance task more tractable I went and purchased an iRobot Roomba 560 ($350) and an iRobot Scooba 330 ($300). The former vacuums the floor while the latter, in theory, washes it.

After a few months of use I can report the Roomba 560 works fantastically well. It can be set to perambulate around in the wee small hours and deals pretty well with long, blond hair. I say “pretty well” because you do have to pick the hair out of the brush mechanism which is why iRobot has the Roomba 532 ($319) with two cleaning bins and two sets of brushes.

I’m not sure that the extra gear would make a difference as hair always gets caught in the brush mechanism, eventually causing so much friction the Roomba gives up and switches off. I had to search online to find out what the problem was and it turns out the solution is to remove the brush bearings and fish out the hair that has wrapped around the axle. Way too fiddly.

But a couple of cool things about Roombas: they automatically redock with their charging station; and they come with a couple of “walls”, little battery-powered boxes that provide an infrared barrier that the Roomba won’t cross, allowing you to keep the ‘bot out of a room or restrict it to certain areas.

The engineering of the Roombas is pretty clever. They use some kind of pseudorandom navigation to maximize the coverage of any size and shape of room and they mostly deal with obstacles and furniture amazingly well. That said, my Roomba does very occasionally get caught under one of my sofas or trapped by the crossbar of one particular chair. In either case it complains audibly then switches itself off. As this is wont to occur in the wee small hours, I get to play find the ‘bot about once per week.

Emptying and cleaning the Roomba is easy and it is always shocking how much crap has been sucked up, even in an ostensibly clean room. Yuck.

The Scooba is another matter entirely. You fill up its tank with water and vinegar, or water and a special cleaning solution and it runs around wetting the floor and sucking up most of the liquid along with the dirt. It does an OK job but every fluid option I’ve tried has resulted in streaks on my wooden floors.

Now, it is true that my floors have a shine that tends to make streaks highly visible and, while there’s no doubt that the Scooba is actually cleaning the floor (the amount of dirt in the onboard tank at the end of a cleaning cycle is, well, seriously gross), the aesthetic results are disappointing and call for a follow-up manual mopping which rather reduces the value of the Scooba.

So it was that two recent robotic floor cleaning products caught my attention and I got them in to test.

The first is the Neato XV-11 ($400) from Neato Robotics, essentially a direct competitor to the iRobot Roomba line.

The XV-11 uses a laser range finder system housed in a small turret on top of the device to map out the room so as to optimize cleaning and augment its obstacle detection and avoidance abilities.

When you start the XV-11 you realize that there’s a surprisingly powerful vacuum system on board … the XV-11 sounds like a miniature Lear jet as it brings its fans up to speed and it is pretty loud in operation.

Like the Roomba, the XV-11 comes with a charging dock that it will usually locate and dock with at the end of its cleaning cycle and, also like the Roomba, you can program what time it starts cleaning.

The XV-11’s physical collision detection system works well for most objects it encounters but the laser turret turns out to be a problem. The device’s shape (rounded on one end) is designed to make it highly maneuverable and allow it to actually get into corners (something that the Roomba’s circular shape can’t do as well).

But here’s the thing; if the XV-11’s body is low enough to pass under, say, a chair’s crossbar there’s no guarantee that the turret will also be low enough to avoid getting caught. As the turret has no collision detection and the XV-11 has some pretty musclely motors, one of two things will happen: Either the XV-11 will stop and complain that it’s trapped and go to sleep or it will drag the chair around. This happens with some bar stools I have; I set the XV-11 going and came back to find the stools had all been dragged around the room after which the XV-11 had got stuck under a sofa.

The XV-11 can be constrained by a “boundary marker”, a strip of plastic that you lay across a doorway or anywhere else you want to keep it from entering. Compared to the iRobot “walls” this is a clumsy solution.

Another problem the XV-11 has in my house is fringed carpets: About half of the time the XV-11 will roll happily over the fringe while the rest of the time it will suck the fringe into its vacuum system and stall out, sulking until it gets freed.

There’s no doubt that the cleaning power of the XV-11 is excellent (the dirt in the hopper even after a short run qualifies as embarrassing), but its inability to avoid getting trapped under the couch and its demented rearranging of my bar stools makes it, at least in my house, a non-starter.

The other device that I tested was the Mint ($250) designed and manufactured by Evolution Robotics. This is a very different design from the other products; it’s essentially an intelligent, powered Swiffer. It is a box with a pad on the front held in place by magnets. You detach the pad, wrap either of the provided microfiber or wet mop cloths or use an actual Swiffer cloth, place the pad back on the body, select a wet or dry cleaning program, and off it goes.

Along with the Mint comes a small box with some sexy blue LEDs that you place in the middle of your room. When the Mint starts it communicates with the box which, using a technology called Northstar, paints an invisible marker on the ceiling for location and navigation, enabling the Mint to travel in straight lines and learn the shape of its environment as well as return to its starting point at the end of the cleaning cycle.

In its wet mop mode the Mint executes a herring bone path, driving forwards and to the right, then rolling partially back and then forwards again to the left, a cycle it then repeats until it has to turn. The results on my floors are amazing! With just water on the mop cloth there’s little streaking and the mop cloth shows just how filthy my dogs make the floor (again, seriously gross).

In dry sweep mode the Mint zips around in straight lines and in both modes it finishes its cleaning run with a remarkably accurate traverse around the room’s perimeter.

So whats wrong with the Mint? First, its battery recharge time is 7 to 10 hours but its run time is only 2 to 3 hours. Second, there’s a handle on the back end of the Mint that’s in the wrong place. When you pick up the device, I would expect it to hang pad up so you could easily replace the cloth and or plug in the charger. With the handle where it is at the back, the pad is facing down. It may sound trivial but after you use the Mint for a while you realize this detail matters.

Finally, the Mint’s charging system requires that you plug the cord from its power supply into a socket on the device’s underside and, given the handle doesn’t orient the Mint to rest on its back end, makes the whole charging process clumsy. The Mint really needs a charging base and the handle relocated so you can pick it up and place it back end down on the charging station.

Despite all of those criticisms I love the Mint. It does its job and the results are excellent.

So looking at all of these ‘bots what impresses me is that they each show off some cool technologies that are impressive and all of the devices work, albeit with varying degrees of success.

So it is that the Roomba gets a 4.5 out of 5, the Scooba, a 2 out of 5, the XV-11 a 2.5 out of 5, and the Mint a 4.5 out of 5. And taking all of these products as a group I’ll give the entire robotic floor cleaning world a 3.5 out of 5. Room for improvement? Sure, but my floors are already looking better.