Dreame's latest robot vacuum concept has slightly terrifying legs that can climb full-size stairs The Cyber X can maneuver steps up to 25 cm tall. The Dreame Cyber X is a prototype for now. Robot vacuum companies are once again trying to outdo each other at CES 2026. This year, Chinese appliance maker Dreame is showing off a prototype of a device that can climb up and down an entire flight of stairs. The concept, called the Cyber X, was previewed last year at IFA in Berlin.

Dreame's robot vacuum with an arm is back at CES 2026 and it can do more than pick up shoes The Cyber 10 Ultra has its own attachments for reaching hard-to-get places. Dreame's Cyber10 Ultra has an arm that can grab objects and clean hard to reach areas. Last year at CES, Dreame showed off a robot vacuum prototype with a mechanical arm . But while we were able to see the arm extend and retract, we didn't see the device, which was described as a prototype at the time, actually grab anything, which was a bit disappointing. This year, though, the company has made its arm-enabled vacuum a reality with the Cyber 10 Ultra.

With pool-mapping capabilities and other smart features, the Dreame Z1 Pro is one of the most intelligent robots I've tested to date. From the start, Dreame's Z1 Pro robotic pool cleaner certainly seems to check off all the boxes. Its features list touts just about everything: The ability to clean floor, walls, and waterline. I'm not sure what the touted "Triple Surround Fusion Perception System" is, but that sounds good, too. I'll start with what I liked the most: After unboxing, I discovered that the 27-pound robot offers one of the most convenient charging systems I've seen to date, thanks to a magnetic charging mechanism that simply snaps onto the device's chassis, with no plugs or rubber gaskets involved--and no need to hoist the robot onto a bulky charging station, either.

CES is known for splashy TV and laptop launches, but in 2025, there were more than a few companies trying to make waves with robot vacuums, too. After years where it seemed like vacuuming, mopping, and self-emptying might be the peak of a robot vacuum's abilities, this year's show introduced home robots with whole new functionality -- and appendages. Here's the robot vacuums that caught our eye at CES 2025, many of which you could invite to clean your living room later this year. Roborock's flagship vacuum for 2025 is the Roborock Saros Z70, a circular robot vacuum on the outside, that hides a retractable "OmniGrip" arm that can pick up and move objects on the inside. The arm is capable of five axis movement and is only able to carry around 300 grams, which makes it better at moving clothes and cables than anything truly heavy.

It turns out that Roborock isn't the only company that brought a robot vacuum with a mechanical arm to CES 2025. Rival company Dreame, which unveiled its stair-climbing robot vacuum earlier in the week, is also working on a robot vacuum with an arm for picking up objects. The device is still a prototype, according to the company, but the as yet unnamed robo vac was on full display at Dreame's CES booth. Considering it's still a prototype, the actual arm looked far more substantial compared to the one on Roborock's Saros Z70. It was much thicker and had a bigger "claw" that looked like it might be able to pick up slightly heavier objects.

Dreame is making a splash at CES this year with a new pool cleaner that's smart enough to gauge the shape of your pool before it dives in for a cleaning, along with a robot vacuum with retractable, obstacle-climbing legs. Slated to ship in March, the Z1 Pro robotic pool cleaner boasts Dreame's PoolSense technology, which intelligently recognizes pool shapes and adjusts its cleaning strategy for optimal results. With 8,000 GPH suction power and dual brush technology, the Z1 Pro promises thorough cleaning, removing debris from floors, walls, and waterlines. The Z1 Pro's LiFi Control System utilizes optical communication to offer precise control over cleaning patterns and paths. Dreame says its horizontal waterline cleaning technology is now 50 percent more efficient than in previous models, while the Auto Poolside Parking feature simplifies retrieval.

Robot vacuums can make cleaning up a lot more convenient but, depending on your home's layout, might not work with every floorplan. Chinese appliance maker Dreame is hoping to address that with its X50 robot vacuum that's able to climb over small stairs without assistance. Though the robot vac has yet to launch in the US, it's already generated some buzz thanks to a teaser posted on YouTube last year (the X50 is already available in China). Now, the 1699 X50 is making its US debut at CES 2025, with Dreame offering an up-close look at its abilities. Dreame says the X50 can handle stairs up to 6cm high (about 2.4 inches), thanks to its "ProLeap System," which relies on a motorized arm underneath the vacuum that's able to swing out and boost it over steps. So while the X50 won't be able to climb up a full-size staircase (the average stair height in a home is between 7 and 8 inches, according to Google), it should be able to maneuver over smaller steps and other obstacles that may trip up other robot vacuums.

Domain generalization (DG) methods aim to develop models that generalize to settings where the test distribution is different from the training data. In this paper, we focus on the challenging problem of multi-source zero shot DG (MDG), where labeled training data from multiple source domains is available but with no access to data from the target domain. A wide range of solutions have been proposed for this problem, including the state-of-the-art multi-domain ensembling approaches. Despite these advances, the na\"ive ERM solution of pooling all source data together and training a single classifier is surprisingly effective on standard benchmarks. In this paper, we hypothesize that, it is important to elucidate the link between pre-specified domain labels and MDG performance, in order to explain this behavior. More specifically, we consider two popular classes of MDG algorithms -- distributional robust optimization (DRO) and multi-domain ensembles, in order to demonstrate how inferring custom domain groups can lead to consistent improvements over the original domain labels that come with the dataset. To this end, we propose (i) Group-DRO++, which incorporates an explicit clustering step to identify custom domains in an existing DRO technique; and (ii) DReaME, which produces effective multi-domain ensembles through implicit domain re-labeling with a novel meta-optimization algorithm. Using empirical studies on multiple standard benchmarks, we show that our variants consistently outperform ERM by significant margins (1.5% - 9%), and produce state-of-the-art MDG performance. Our code can be found at https://github.com/kowshikthopalli/DREAME