When technology is at its best and focused on improving our lives, like a great-sounding pair of wireless headphones or a really good computer mouse, it can be indispensable. But when the companies making our gadgets drift away from the user experience and start checking off boxes and publishing bigger numbers, the basics can get lost in the mix.
That’s exactly what’s happening with the Dreame Aqua10 Ultra Roller, a premium robot vacuum with perhaps too many ideas. Which is not to say the ideas are bad. In fact, the Aqua10 Ultra Roller is gadget-y in ways that I always hoped robots would be when I was a kid in the ‘90s, all whirring motors and parts that pop out of its body to do things. Its mop roller juts out of its side for better edge mopping, the periscope-like sensor array sinks into the body to let it go under things, and little legs let it climb over obstacles and onto slightly higher surfaces. I love that stuff, and it all seems to work just fine.
Dreame Aqua10 Ultra Roller
The Dreame Aqua10 Ultra Roller is an expensive disappointment.
- Lots of clever robotic features
- Solid mopping performance
- Self-maintaining mop roller
- Tangle-free roller brush
- Mediocre vacuuming
- Very buggy navigation
- Poor battery life
- Too expensive
- Unintuitive, complicated app
But things haven’t come together for the Aqua10 elsewhere. And unfortunately, it was the robot vacuum basics—navigating my home, sucking up debris, and managing its own charge—where the Aqua10 failed the hardest while I was testing it. I think Dreame could fix these problems with a few software updates, but as things stand, I’d be sorely disappointed if I’d paid $1,600 for a product that has so much trouble with these standard tasks.
All the bells, all the whistles
© Wes Davis / Gizmodo
The Dreame Aqua10 Ultra Roller is a big white circle that stands tall, at 3.84 inches, or 4.72 inches with its sensor array extended. Where it differs from a lot of robot vacuums is all those robotic parts serving different purposes. It’s fun watching the sensor array lower until it’s flush with the body to get under my couch, and it’s satisfying to see the mop roller slide out to follow the contours of furniture. Same goes for the side brush, when it pops out to slap dust out of corners its big disc shape would otherwise make difficult to reach.
Tucked into the Aqua10’s side wheel wells is a pair of stubby legs, with wheels on the ends, that unfold and jut downward to shove the robot up at an angle. (In this mode, it sort of resembles the Wheelers from Return to Oz, except not terrifying.) I don’t have any transitions that actually require the Aqua10 to use this feature—though it did decide, twice, that a dining room rug in my home called for it—so I stacked up some shelving wood I had lying around. I couldn’t get to exactly the 1.65 inches that Dreame says it can handle at once, but the robot climbed onto a 1.5-inch-tall stack just fine. It’s no stair climber, but I have no worries about this robot vacuum balking at thicker carpets.
© GIF by Wes Davis / Gizmodo
You can adjust how each of the robotic parts behaves if you’re willing to explore the labyrinthine depths of the Dreamehome app. Here, you’ll find pages and pages of menus, toggles, and sliders that seemingly let you tweak everything about the robot. Want to set the mop roller never to extend, or to do so only once a week? You can do that. Want to set the robot to clean every night, but to avoid mopping one specific room on Tuesdays and Thursdays? Also possible. You can set suction power; or decide if you want the robot to vacuum and mop room-by-room or vacuum every room, then mop every room; or set it only to mop in the direction of a hardwood floor’s planks to avoid roughing up the long edges.
Neat stuff, and it’s nice to have options, but wow is it a lot, and not to mention arranged in a way that I think will turn many folks into the Homer Simpson backing into shrubbery meme. That could be mitigated were more power user-oriented features tucked one layer deeper than the more straightforward ones like scheduling and cleaning history, which ought to be right on the Aqua10 landing page and aren’t.
© Wes Davis / Gizmodo
Also, some options just feel like padding. For instance, on one page, you can tell the robot not to mop the carpet. That should just be standard behavior—I shouldn’t need to forbid rug-mopping! Others don’t appear to do anything at all, including many parts of a suite of “Pet Care” features. With Pet Care turned on, the Aqua10 is supposed to do more intense cleaning around things like pet dishes and litter boxes. It was hard to evaluate this around my dogs’ dishes—mainly because they just aren’t messy eaters—but the Aqua10 seemed to do the opposite of intense cleaning around our cat’s litter box, leaving a lot of debris behind. “Pet Moments” identifies your pets and takes pictures of them, although it only captured two images—one still and one GIF—of mine during my time with it. Dreame told Gizmodo it’s possible my pets were just “camera-shy,” which they weren’t; they’re very used to robot vacuum hijinks and seem to almost relish being in the way. Perhaps the Aqua10 looked upon them and found them wanting (in which case, sorry to Dreame’s algorithms, but we are in disagreement).
Other features just aren’t labeled clearly, a common feature of apps made by non-English-speaking developers. Take the toggle for Collision-Avoidance Mode. I would assume that turning this off would have the device barreling into walls and furniture, but that’s not really what it does. It’s more like switching it from a vacuum that’s very cautious not to touch walls to one that isn’t quite as careful.
© Wes Davis / Gizmodo
Finally, there’s not a whiff of support for the universal smart home platform Matter, despite the fact that it’s promised on the Aqua10’s product page. Dreame told Gizmodo that Matter support is coming by the end of November with a software update. That’s all good, but why advertise it as being Matter-compliant if it’s not yet? Until that update, Aqua10 owners will have to make do with Amazon Alexa or Google Home integration, basic automations using the Apple Shortcuts app, or the robot’s built-in voice assistant, which is clunky at best and requires a lot of rote memorization of pre-programmed commands.
It’s possible that Dreame bit off more than it could chew in time for launch. But if these features weren’t ready yet, the least the company could’ve done was mark them “coming soon” or “beta.” Otherwise, it just feels like all these toggles are just there to make the robot seem more featureful than it actually is.
Aggressive mediocrity
© Wes Davis / Gizmodo
I was deeply annoyed once I actually started using the Aqua10. Before we get to why, let’s start with what this robot does well. Thing number one is all of the robotic stuff. Apart from the robotic legs doing their thing when I tried testing it, the extending mop roller is really good at following the contours of things like furniture while mopping, slipping in and out of the Aqua10’s side as it drives alongside them. And it was cool watching it sink its sensor cluster down into its body to go under furniture, then raise it again as it exited. Its object recognition is very solid—it definitely still tried to vacuum up things like screws and marbles, but the Aqua10 would not be fooled into sucking up socks or towels I placed in its way, nor a pile of coffee grounds I plopped down as a makeshift pet poop simulation.
Mopping performance was solid, although not as good as Dreame’s marketing materials would have you believe (imagine that!). Despite all of its features—a robotic mop roller, which self-rinses during cleans and gets a heated bath of sorts when docked; a software slider in the Dreamhome app to set how wet you want the roller during cleans; options to set higher or lower downward mopping pressure; and the ability to tweak how tightly it overlaps its cleaning passes—the Aqua10 is good for regular maintenance mopping and not much more. It left behind streaks of ketchup when I squirted a patch of my floor with the stuff. It didn’t make a dent in a dried mystery stain in my dining room. In both cases, the roughly $500-cheaper Matic robot I recently reviewed did a better job, totally clearing the ketchup as well as the exact stain the Aqua10 failed to clean. On the plus side, the Aqua10 never mopped any of my rugs—its ability to discern between carpets and hard floors was spot-on, at least in my house.
© Wes Davis / Gizmodo
Before and after mopping runs, the Aqua10 Ultra Roller’s dock—a complicated, mini-fridge-sized piece of machinery—goes to work. Behind a panel in the front are the floor cleaning solutions and dustbag, and beneath the lid on top, two large tubs for clean and dirty water. You’ll hear humming, gurgling, and spitting from the dock as it pumps water to and from the robot in preparation for a clean. During this, the Aqua10 sprays heated water over its mop roller, which turns, scrubbing itself on a recessed, textured plate in the dock. This whole process lasts for several minutes as the Aqua10’s built-in speaker announces, each step of the way, what it’s doing. (If you find that as annoying as I do, you can turn the voice down; all the way to zero, if you want.) All of this isn’t too loud, per se, it’s just an aggressive amount of activity for something that just does an okay job at mopping.
© Wes Davis / Gizmodo © Wes Davis / Gizmodo
While we’re on the subject of noise: When the Aqua10’s dock auto-empties the robot’s dustbin, it’s about as blaring as the Eufy L35+ Hybrid I own. Which is to say, it’s startling if you’re in the same room and didn’t expect it, and I would definitely restrict when it can do that, using the app’s Do Not Disturb schedule.
As for vacuuming, or what is really job number one of these devices, the Aqua 10 wasn’t a lot better than my almost-three-year-old and poorly maintained Roomba J7. It missed a lot more than I would’ve expected during nightly cleans, leaving behind little scraps of paper or small rocks that had been tracked into my house. The same goes for spot cleans; the Aqua10 would spread bits of dirt with its side brush, flinging them out of range before missing them later.
© Wes Davis / Gizmodo
When I ran the Aqua10 on my basement rugs, which don’t get vacuumed as often as the rest of my home, it did this weird thing where it pooped out balled-up hair and string after they were caught in its anti-tangle brush roller, then left them behind when it finished cleaning. Sure, just about every robot vacuum misses things, and picking balls of hair up by hand is definitely better than having to painstakingly cut them off of a roller like with the Roomba J7, but if I’m paying $1,600 for a robot, I don’t want to do nearly this much cleaning up after it. When I flipped the robot over and looked under its dual rollers after a week of testing, I saw a clog forming where built-up hair and string was partially blocking the area around the suction hole, which I thought could be why the Aqua10 was leaving so much behind on its cleans, but it didn’t do better when I ran it again after clearing the blockage.
The worst part of the whole experience was that I couldn’t count on the Aqua10 Ultra Roller to finish a clean without babysitting it. It needed to recharge itself after cleaning between 190 sq. ft. and 300 sq. ft. of my house every night—battery life is definitely an issue—and so would stop working and go look for the dock. More than half the time it didn’t make it there. Most of those times, it found the dock but just didn’t manage to park in it—why it didn’t was unclear, and it would dock just fine when I found it in the morning and pressed the Aqua10’s physical home button. Once, it drove around to the hallway behind where the dock is and sat there until it ran out of battery. Every robot vacuum I’ve ever owned or tried has similar problems, but usually only occasionally; this was almost every night, even after remapping the floor and taking care to move the chairs in my dining room, where the dock lived, so it had plenty of runway.
I watched the Aqua10 roll over and fail to vacuum these things moments before this picture. © Wes Davis / Gizmodo
On one of the final nights of testing, I sat and watched the Aqua10 clean, then timed it as it spun around and around in front of its dock at the end, pausing to move a bit here and there before spinning in more circles. It took more than six minutes for it to settle on an approach vector and park to recharge. It was like watching a bugged-out Skyrim NPC. This was after it not only drove over several bits of dirt and paper throughout my home without collecting them, but also spit out new pieces of garbage in my dining room that it got from who-knows-where.
Privacy is a concern
Robot vacuums have an unusually personal kind of access to us that other products do not. If their manufacturers want, they can gather details about the layout of your home, your habits throughout the day, how your space changes as time goes on, how many people live in or visit your home, and even how much dirt and dust you generate. Their cameras are pointed at you and your home from all kinds of angles, and some, like the Aqua10, even have microphones to power voice assistants. I’m not saying Dreame is abusing this access, but I am saying that when products like this require an internet connection to work, they also require a massive level of trust. Robot vacuum companies probably can’t glean as detailed a picture of your life as a smartphone manufacturer can, but the data points could still be valuable as marketers (or government agencies) seek to build a more comprehensive picture of who you are and what you like to do. It’s not something we all want to think too deeply about, and many of us are resigned to it at this point. But it’s still worth looking at what your robot vacuum and its associated software might be doing behind the scenes.
And just so you know, there is a lot of extra chatter coming from the Dreamhome app, especially the first few days after I set up the Aqua10. My iOS App Privacy Report indicated that Dreamehome had, at one point, contacted 185 different domains in the seven days prior. Many were Dreame’s own domains, but others belonged to companies like Facebook, Baidu, and Google, and some—one of which the app had contacted more than 500 times in just a few days—are completely unnamed. (When I looked up the one in question, it appeared to be Alibaba’s cloud compute service.) The next most-active platform, YouTube, had pinged 173 mostly Google-owned domains, one of which it reached nearly 400 times. For a more direct comparison, the iRobot app had reached out to 37 domains, including a handful of what looked like trackers, and the Matic app had contacted exactly one web address: the local IP address of the Matic robot.
© Screenshots by Wes Davis / Gizmodo
Like a lot of people, I’ve come to accept that there is a data privacy trade-off when it comes to using certain devices. I don’t like it, but it’s nearly impossible to use modern technology without accepting it. And as a gadget reviewer, I give my data to a lot of different apps in the course of my work. But I am highly suspicious of any app that’s contacting that many domains. Dreame’s spokesperson told me some of the app’s features are “the result of collaboration with external partners,” who provide it with “a ready-made system that includes a variety of embedded third-party tools,” like statistical code and video players. These tools, they said, “make calls to their own respective servers, which accounts for the high volume of network connections that you have observed.”
That’s all well and good, and the app seemed to settle down after the first few days of testing to just contacting 53 domains in the previous 7 days as of this writing. And yet, even if that’s mostly just to support the app’s features, it’s still a lot of opaque outgoing communication from a single smart home appliance’s app.
So busy, now!
In the movie The Fifth Element, Gary Oldman’s character (“Jean-Baptiste. Emmanuel. Zorg.”) pushes a drinking glass off his desk to illustrate his villainous view that by destroying things, he gives life the opportunity to flourish. As the glass shatters, several robots parade out from a hidden wall compartment to clean up the mess. His idea is that these robots were created by hundreds of people who are able to keep feeding their families and prosper through this sort of continuous destruction.
It’s a scene I kept thinking about while testing this product, and it’s not Zorg’s bullshit ethical posturing that’s been on my mind. It’s something else he says: “Look at all these little things. So busy, now! Notice how each one is useful.” As he’s saying this, two robots with flashing lights cordon off the area, one sweeps up the mess, another sprays liquid onto the ground, and a final one spins about, mopping.
I don’t just bring this up in hopes that someone will validate my taste in movies. It also helps me make a point: for whatever reason, the filmmakers assumed that the best approach to floor-cleaning robots is to make them narrow-purpose devices. I don’t think you have to go as far as to have an individual robot for spraying floor cleaner, but as I wrote this review, I kept thinking about how my favorite gadgets are often the ones focused on being really good at a small number of things. Maybe it’s because if something is made to accomplish one or two tasks, it’s unforgivable if it sucks at those. If it’s made to do a whole bunch of things, it can get away with doing some of them poorly. It feels like Dreame is almost counting on that possibility.
If there’s a bright side, it’s that most of the Aqua10 Ultra Roller’s problems seem like the sort that can, and hopefully will, be fixed with software updates the way a few good patches can turn an unfinished, bad video game good. But it’s one thing to spend between $60 and $70 (or, sigh, $80) on a glitchy, unfinished game. It’s another entirely to drop several hundreds or more on a robot vacuum that can’t appreciably out-vacuum my dusty old Roomba J7, is missing advertised features like Matter support, and has so much trouble finishing a clean. At the end of the day, the only thing the Aqua10 does reasonably well right now is mopping. And, uh, who wants to pay $1,600 for that? Until and unless its issues are fixed, it doesn’t seem worth it. Just buy a mop.
