Roborock Saros Z70 review: More slip than grip

I am the target audience for a robot vacuum cleaner with an arm. Between my partner, myself, and our dog, there are always socks and toys littering the floor. The idea of a robot picking these up wasn’t just a novelty for me, it was a game changer. As such I have eagerly awaited my chance to review Roborock’s Saros Z70 - the first robot vacuum in the arms race - but by-and-large it has been a let down.

Not only has the robotic arm proven to be inconsistent and unreliable, but the focus given to it means other parts of the vacuum have suffered. While Roborock’s premium robovacs usually set a performance standard, this one has struggled with tasks its predecessors glide through with ease. It doesn’t navigate as well as others in the family, it is more prone to blockages, its debris clearance rate is significantly lower, and it doesn’t even make up for it by picking up my dog’s toys.

Limitations and disappointments are par for the course when new technology is introduced. The first iPhone couldn’t copy and paste, the first foldable phones were too fragile, and the first robot arm on a consumer level vacuum needs some work. When I spoke to Roborock last year, their Marketing Director told me the brand wanted to “do the right thing.” To build the best products and to make a name for themselves based on that. Up until now, I believed that, but looking at the Z70 I’m starting to think that approach may have changed. While there is glory in being the first brand to put out a product, there is more in being the first to do it well, and when it comes to the Saros Z70, another year of development could have made all the difference.

As it stands, the Z70’s robotic arm isn’t much more than an expensive gimmick. Although it is supposed to be able to pick up, and sort lowline shoes, balls of paper, and socks, its hit rate is abysmal. In the weeks I have been testing the Saros, it has managed to successfully pick up two balls of paper, and deliver one of them to the designated sorting bin. I would need a couple extra arms myself to count how many times it has overlooked other supported items, and another one for all the times it failed to grip them.

The gripping mechanism itself is reminiscent of an arcade claw machine. While it can technically grab hold, and transport items, the claw is more slip than grip. It does keep me coming back to try again, but the extra $1000 this robotic arm sets you back is a much higher asking price than another $2 on a Timezone card. Robotic arm support also isn’t available during targeted zone or room cleaning so you’ll need to send it out for a full clean if you want to try your luck.

Even so, the rest of the vacuum’s performance drops when the arm is in use. Otherwise solid navigation and depth perception is noticeably worse when the arm is up, whether it is holding something or not. Without the claw enabled, the Z70 manouevers to a specific spot on the map with no issues but when it is (or thinks it is) gripping something, it falls short by 20-30cm. Ironically these navigation issues make the robovac nearly incapable of getting the item from its grasp into the designated sorting bin.

Roborock told Reviews there are plenty of OTA updates coming that will make the arm smarter and more capable over time. In June it will gain sport shoe support, those navigational issues are getting a fix, and in July the manufacturer is focusing on accuracy. That is all well and good to hear as a reviewer, but it doesn’t hold much solace for consumers who have forked out $4000 for the device. Those of us embroiled in the tech space know that innovation takes time, and that by-and-large new leaps rarely land on the first go. It is easier to be patient, and wait for these improvements and upgrades when you’ve got a room full of products waiting to be reviewed, but for real-life customers the disappointment is going to be far greater.

Of course the robot arm is the real draw card for the Saros Z70, but it is, at its core, a robot vacuum cleaner and mop. I’ve already mentioned it, but the performance on these core features are significantly poorer than what I’ve grown to expect from Roborock devices. While other recent releases like the Qrevo Curv and the Saros 10 have high clearance rates for debris and wet messes, the Z70 struggles with both.

Small and medium debris has only a 70% clearance rate, with plenty of the remainder getting spread out during the cleaning process. Although other Roborock vacuums have easily sucked up the sprinkles and oats I use for testing, I found the Z70 pushed and spread a lot of these out, and ended up leaving a noticeable amount of debris after cleaning. In part I put this down to the new roller. It uses a comb like design to minimise hair and fur tangles, but the small gaps make it harder for other debris to get pulled into the dustbin. I noticed my testing dander getting stuck in the small spaces between the comb teeth before it would dislodge later in the cleaning process. Larger debris, which ideally the arm would be able to handle, and small hard tidbits like sewing needles instead tend to get stuck in the teeth and make an absolutely horrific noise. On the up side, fur and hair haven’t been an issue, so that’s something. Overall, the vacuuming performance is not up to scratch for this price point.

The mopping is a similar story, but for different reasons. The actual clearance rate for wet and sticky messes is around the same as most other dual-mopping pad robot mops - it will pick up almost everything but leave a sticky residue in its wake. Whether it actually mops or not is the issue. The mop heads automatically detach at the base station to avoid dragging unnecessary moisture and dirt around the floors, but it doesn’t always pick them back up when it is time to mop. More times than not the Saros Z70 has simply not run a mop cycle, even though it is set up to do so. It vacuums, it claws, it empties, and it stays at the base station. Regardless as to how good the mopping performance is, if it doesn’t reliably carry it out then it defeats the purpose.

There is something to be said about being the first to bring one of these new tech leaps to market. Now, and forever Roborock will be the first robovac with an inbuilt arm. No one can take that away from them, but its performance at release will forever be marked alongside it. On a whole we tend to remember the quirks and the failures for far longer than we remember the successes. When Apple launched Apple Maps a slew of issues followed, and even today - when Apple Maps is actually a good product - performance assumptions plague it. First impressions count, and I think Roborock may have goofed it here.

From my testing I know Roborock to be the Apple of the sector. It consistently churns out excellent, and reliable products with thoughtful innovations that are shiny and polished by the time they hit shelves. They don’t compromise on the user experience just to make it to market. Or, they didn’t until now. Apple isn’t the perfect example of this ideal anymore, but for the most part Apple’s philosophy hasn’t been about being first, it has been about being the best. We saw it with the Vision Pro, and even though it hasn’t gripped the market it is widely considered the best virtual and augmented reality device on the market. I can’t help but wonder what Roborock’s robotic arm could have been if it was given the same treatment. If all the effort put into these OTA updates came before the fact rather than after it. Maybe this review would read differently, maybe I’d be holding the robot’s hand rather than giving it any other gesture, but we’ll never know.