Common design flaws in robot vacuum cleaners

I’ve been using 2 robot vacuum cleaners (Midea MR06 and Rowenta Series 60 Anti-Allergy) for a while now, one particularly in the room with computer and bed where dust just never seems to be really gone. I’ve also got quite some experience with entire range of iRobot models and watched quite few tests online to really understand robots quite well. And here are some common design flaws that I want to point out so maybe vendors will fix or improve them. Or just for users to know what to look for when buying one. I’m mostly going to point out the above models, but I’ve also seen a lot about Dyson 360 as well as bunch of “no name” Chinese robots that have the following problems… I hope robot vendors will read this as well as users when looking for new robot. Just decide if mentioned design will be a problem for you or not and then narrow the choices. The list will probably get really long, but I’ll make categories so you can easily jump to things that may interest you more.

Ok, lets go. 🙂

Chassis design (Height)

A lot of robots are pretty standard size, give or take few millimeters. There are 2 models I’m aware of that really stand out. One is Rowenta Explorer Series 60 (RR74xx) models that are super slim at insane 6cm of height only. And at the other extreme, Dyson 360 Heurist at 12cm is one of the tallest robots out there. While being lowest may be beneficial in some cases where you want it to go under, it’s sometimes also a curse. My Rowenta Series 60 can go under my 4 wheel cabinet just enough so it cleans a bit underneath it where other robot bumped into its edge right away, but it also managed to drive itself underneath my office chair and managed to get stuck between 5 wheels of the chair and error out there because it just couldn’t find a way out. Now I need to park the chair and spin the base with wheels so they don’t leave enough space for robot to roll underneath again.

Chassis design (Rubber bumpers)

The biggest design flaw I’ve noticed is poor protective rubber bumper placement. Or lack of it entirely. Roomba S9+, the most expensive robot from iRobot doesn’t have any. Yet it uses front face bumper to register some collisions. You have no idea how ugly robot looks after it has some range on its clock. Entire front glossy as well as matte plastic full of scratches and dents. It’s not a pleasant sight. Dyson 360 has the same problem. Being tall and without any rubber bumper, it’ll look all messed up and worn out over time. These may have advance sensors and cameras, but they still hit into things. Not to mention it will eventually also damage furniture and equipment in your home with repeating collisions.

Some are a bit better by having a silicone/rubber bumpers in the front in a form of a strip, but they are often either placed way too low or way too high, so it keeps hitting things with plastic. Why is it so hard to place two rubber strips, one at the lowest point and another at highest? Chance of getting something in between becomes very unlikely. I’ve added a strip of self adhesive window gasket on one robot at the highest point and then it stopped hitting things with chassis directly. Looks a bit ghetto if you look close, but why wasn’t this done properly in the factory?

Docking station avoidance radius

One of biggest issues a lot of robots have is docking station avoidance algorithms. Midea robot often literally ran over its own docking station when cleaning. Never got stuck, but it makes funny noises. And it cleaned dust of the station which was useful. Rowenta, avoids it in a moderate arc. Similar with iRobot. But I’ve seen some that avoid it in 2 meter radius which is just ridiculous. I have my robot under the desk and such huge avoidance radius would mean it would never ever even clean underneath the desk.

Docking station finding when returning

It’s funny how some robots have issues returning home. Usually those with cameras do just fine because they have a visual representation of the room and where they started, but a lot of camera-less systems seem to just get lost and stuck somewhere after they finish. Rowenta Series 60 for example is often acting like a dork. It has the dock in sight, like 1,5m away yet it bumps, rotates and drives in opposite direction. It usually finds way back, but it’s just funny how it’s almost there, but doesn’t see it. Midea on the other hand looked like it had a rough idea where to go for docking and then looked up the dock when in range. When it couldn’t, it followed the walls until it found the dock. Because you usually place the dock somewhere against the wall. Which is a neat trick. At worst it’ll do a larger circle around room, but it’ll find it.

Cleaning start condition free or from dock

Apparently some robots are not aware of the starting condition. Meaning, if you take it off the dock and for example carry it to a different floor and send it to work there, it’ll just fruitlessly try to find the docking station after cleaning is done. I don’t have enough info how common this is and at least for Midea it seems to be aware whether it starts from the dock and when it doesn’t. When started manually off the dock it always instantly stops and says “Cleaning completed” when it’s done. But when it starts from the dock, it tries to return to it. Which is a very basic but neat thing.

Side brush design

There are several quite common design flaws when it comes to side brushes. And it’s pretty amazing or shall I say shocking that so many vendors mess this up.

Most common problem is that side brushes spin way too quickly, flicking debris around. This especially becomes problematic with camera navigation which is very systematic and anything it flicks into the area that was marked as “already cleaned”, it’ll never return there to pick it up.

Another issue are designs with single side brush, most commonly used by Roomba robots. Sure it has logic so it always does the wall following cycle with the correct side, but the issue with single brush is that only one side is sweeping debris and dirt inwards toward the main roller brush. Which often means the wheel on side without the brush is driving on dirt, smearing it around. Where dual brushes create a super wide zone that’s moving dirt inwards where main roller brush picks it up and collects in the dust bin. And dual side brushes in most cases also sweep against the wheels, cleaning them as well as brushes itself.

Third issue are very short side brushes. Again, most noticeable on Roomba S9 series. Single, very short brush. Not only it just twists in a really weird way on carpets, its reach underneath furniture and against walls is quite noticeable as it often leaves debris near walls and just doesn’t clean dust under furniture where it’s only few millimeters of gap between furniture and ground.

Main roller brush

There are generally 2 different designs. Dual brush design which is patented by iRobot and single brush design used by everyone else. Roomba dual brushes rotate in opposing directions, inwards. This can help grabbing larger objects and using rubber brushes without bristles helps with pet hair and just hair in general coz it doesn’t get tangled up on roller brushes. Downside is they also tend to flick things instead of picking them up. Also makes robot easier to climb on carpets.

Other robots, well at least I can fully confirm for Midea and Rowenta S60 use a rubber strip right behind the roller brush, acting like a scoop. It’ll never flick anything past it because the rubber strip is all the way to the ground and brush is rotating against it, brushing up any and all debris into the bin. Downside is that these scooping rubber strips tend to catch on carpet edges, causing robot to have problems climbing thicker carpets.

Another funny design is yet again with Dyson 360 where wheels (well, “Tank Tracks”) are actually in front of the roller brush as it’s driving around. It’s funny because it means its own wheels always go across all the dirt, increasing chance of things tangling up into drivetrain as well as smearing dirt or rubbing it into surfaces instead of picking it up first and then driving own wheels across clean surface. All others have roller burshes in front and those that do not have side brushes that sweep in front of wheels which neatly solves that “issue”.

Dust bin design

Of all robots I’ve seen and even used, Midea MR06 has the best dust bin design I’ve seen. Period. Full stop. It has a lot of space, one way ingest hatch, large prefilter, large main filter, easy access to all its parts, easy access to main filter and the dust bin can easily be taken out of robot whether on dock or not. Benefit of such design is super easy maintenance and super high efficiency. One way hatch ensures spiders and bugs that it sucks into the bin can’t escape. Which is super neat. Large prefilter and main filter ensure robot can pick up huge amounts of dust without ever getting it fully covered in dust. Usually entire dust bin is absolutely filled with dust chunks before it even clogs the prefilter. Design on MR06 is so good it once managed to gobble up entire really thick crayon that my nephew dropped under the sofa. When bin was already almost full of dust. That’s super impressive.

Rowenta S60 for example has easy to the dust bin itself, but has no one way hatch on the dust bin entrance which is rather narrow so it wouldn’t ever be able to pick up that crayon, prefilter is attached to main filter and both are tiny. Which means dust will build up on prefilter and clog both prefilter and main filter, way before dust bin is even full. Roombas with microfilter have these weird boxes with this bizarre plastic trap hatch that doesn’t really serve any purpose and higher end models like E5 and E6 have dust bins that have a very rough prefilter with no way of getting inside the dust bin for easy maintenance. Both Rowenta and iRobot are huge companies and you’d expect them to do better than this. I’m still looking for a way to make a larger prefilter for my S60. Would be relatively simple if I had a 3D printer, but I’ll think of something. Currently I’ve added standard vacuum cleaner motor filter of same size as main filter which seems to help a lot. But it’s still same size as main filter. I want to make a prefilter of same size as the entire width of the dust bin and make it removable for easy cleaning. That should help it hugely.

And lastly, what annoyed me the most with Dyson 360, the orientation of the dust bin when on dock. The dust bin on the robot is facing wall, not the user. Which makes it very clumsy to take out of the robot.

Wet cleaning attachments are a gimmick

Now to the wet cleaning. My Rowenta S60 happens to have an attachment for wet cleaning, a water tank with microfiber pad underneath. Water slowly drips onto the microfiber pad ensuring it’s wet entire time as it glides across surfaces. I’m not gonna say it’s entirely useless as it did wipe some dirt that didn’t get picked up by brushes and suction, but my god I shouldn’t have looked inside the robot. Wheels were all wet and smeared in wet dust as it was driving across already cleaned (and thus wet) surfaces, side brushes were wet, there was wet dust smeared all over the main roller brush, entire brush compartment, duct from the brush into the dust bin and even inside the dust bin. It’s a maintenance nightmare! Used it once when I bought it and never again.

I have to give credit to iRobot here, but for different approach. iRobot allows robot pairing. First Roomba vacuums the rooms and when it’s done, it sends signal to Braava robots which do the wet cleaning afterwards. This way Roomba will never drive across wet surfaces and smear wet dust all over its internals and Braava, the wet cleaner will really clean only the dirt that couldn’t be vacuumed regular way. Yes, it’s way more expensive, but it’s really the only proper way of doing it.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s