Roomba j7 + in a practical test: This vacuum robot should no longer knock over anything

Roomba j7
More than just sucking: This robot should recognize obstacles and not knock over anything

Like its predecessors, the j7 can also be controlled via Alexa

© PR

The new Roomba has an HD camera, so it recognizes an obstacle and its AI understands the peculiarities of the home. It works quite well for the beginning and the system gets better and better over time.

The new Roomba j7 wants to set new standards with an HD camera and even smarter navigation. The practical test checks whether this also works. The new j7 should be more cautious and smarter. If you unpack the device, the vacuum cleaner is less noticeable than the suction station, which deposits the dirt from the vacuum cleaner in a bag. It is included in the more expensive j7 + devices. The station is a little wider and less high than the previous one. This makes it possible to make them disappear visually under a console or table. With a small leather bow, the box actually looks more homely. Another plus compared to cheap devices: The vacuum cleaner is partially parked below the station, which means that the floor space is smaller.

Technically based on the i7

The brand new one j7 looks essentially like the older i7. Motors, brushes, dust compartment – they are all identical. However, the piezo sensor of the old device was dispensed with. What is new is an HD camera on the front and a small headlight. For the result it is better if the robot works in light. It looks funnier when it draws its tracks behind its light cone in the dark at night.

To be smart – that’s what many robots promise these days – but how smart is the Roomba really? Laboratory tests provide little information about this because the robots only have to find their way in a small test room and not in a complex floor plan. Functions such as the Roomba’s smart map are not queried in this way. The main focus in the development of the Roomba robots is that the machine can find its way back to the station as often as possible without human help. This makes the j7 work cautiously. To do this, he drives through a room more often. And tends to occasionally switch off the main brushes on carpets with very long fibers. As a result, he takes longer, but the advantage is that he does not run the risk of getting caught in some idiots. A plus compared to other vacuum cleaners.

Flood of functions

To define a complex floor plan for the first time, the j7 needs a few days – even if you allow it to do a training run for mapping. Once the map has been designed, the rooms can be named, the robot can be assigned individual tasks and restricted areas can be assigned. There may be customers who take advantage of the full functionality, but many will not. The whole software cleverness still fails because the doors are locked. When vacuuming in the bathroom or bedroom, a door usually has to be opened. We then carry the robot straight to the scene instead of giving it the order to follow suit.

Recognizes a lot but not everything

But now to the essentials: The j7 is equipped with a high-resolution camera. The so-called PrecisionVision navigation technology is not only used for navigation, this camera is supposed to recognize and identify objects with the help of artificial intelligence (AI). That sounds cool, but it also means in plain language: The system learns. After a while, it becomes wiser than at the beginning. If it encounters unknown objects that tend to be critical, it includes a photo of them in the map. The owner can now decide what to do here. For example, whether it is a temporary obstacle (sneakers with laces) or a permanent source of danger (tangled cables under the desk). How well does the detection work now? At first only obstacles are recognized. It does not cover very small, flat objects such as a coin. So the system will spare the Lego brick. In the intricately designed and by no means perfectly tidy apartment in which the j7 was used, the detection worked very well. Not once did the machine choke on shoelaces or cables, when a mishap happened, then it ate at small objects like a coin or a cable tie.

Getting smarter and smarter

When working, the camera recognizes objects as obstacles and drives around them, Roomba promises. That works well, too, but not one hundred percent. And even with the much-invoked “dog poo”, the results should be “mixed”. As long as it’s a handsome pile on light-colored tiles, the j7 will be able to identify it. If a liquid flatbread lurks on the robot on a dark background, it will hardly work. Basically it is an AI-based system, so it should get better and better through experience and regular software updates. What you get now is only the first stage of development. And then there are occasions. But one of the advantages of the Roomba vacuum cleaner is that older series are also regularly supplied with updates and increased functions. Just like you can easily buy replacement parts and batteries.

Camera recognition is the future at Roomba

Is the j7 + worth it? The price including suction station is a proud 999 euros – but it is cheaper for promotions. The Roomba vacuums always score with a very good suction power. You do not work with a brush, but with two rubber rollers rotating in opposite directions. So does the j7. For him, more cautious behavior means that he tends to take longer to complete a task than less intelligent machines. It is much more important that the machine doesn’t get stranded somewhere and bite into curtains or cables. The main feature, the HD camera with object recognition, certainly does not work perfectly yet, but it works quite well. In any case, the future at iRobot is based on the camera, object recognition and PrecisionVision navigation technology. So if you are considering buying a more expensive device from the market leader, it should be a j7 or a j7 + – because the further development of the software will be linked to the camera. With all too seductive offers, you unfortunately have to be careful of fake shops.

You might also be interested in:

This article contains so-called affiliate links. Further information are available here.

source site-5