Ecovacs DEEBOT OZMO T8 AIVI – AI floor cleaning robot vacuum
The Ecovacs DEEBOT OZMO T8 AIVI (Artificial Intelligence and Visual Interpretation) is its latest robot vacuum. The OZMO part of the name means it can mop as well. Too good to be true? Well, it is a cleaning delight and lives up to its claims of a better-looking home.
We have been testing the Ecovacs DEEBOT OZMO T8 AIVI for the past few weeks. Conclusion – it is a formidable robovac and mop in a single unit.
The Ecovacs DEEBOT OZMO T8 AIVI is a next-gen (2020) robovac. It uses AIVI for object detection and avoidance. If you leave shoes, rugs, electrical cables etc., in its path it can avoid them and let you know via a map what to fix next time.
We can’t verify Ecovacs claims of 200% faster recognition and 60% less entanglement, but we are prepared to say that it is pretty good – at least as good as other next-gen robovacs we have tested.
Ecovacs DEEBOT OZMO T8 AIVI (Model DBX11-11)
We have tested robovacs from LG, Ecovacs Deebot, iRobot, Samsung, Roomba and more. One thing is clear – the technology has improved over the past year. As a result, it makes it harder to stand out from the 2020 pack. Throughout the test, we use the terms FAIL, PASS (meets our expectations for this price/style) and EXCEED against the various paradigms. You will start to see why this DEEBOT is no dumbot.
The standard test is three carpeted bedrooms, large open-plan timber floor living space, two tiled bathrooms, tiled entry foyer and stairs, and a sizeable shaggy feature rug. We cannot vouch for all surfaces or styles of homes.
What is a Ecovacs DEEBOT OZMO T8 AIVI?
First, it is a robot vacuum cleaner based on the ’round’ design. It is 353mm round and 94mm high. The front outer rim is a collision bumper when AI cannot avoid an obstacle or encounters newly placed furniture. Gone are the days of bumping into everything.
It has a charging base 20V/1A/20W docking station CH1822 (so you need to find a suitable place with 240V power).
It includes a removable, passive mop with a friction cleaning cloth plate that can use DEEBOT washable microfibre cleaning cloth (one included) or a one-use disposable cloth (box of five included). This is purely to lightly mop or dry dust rather than to thoroughly clean and polish floors.
And finally, it has the OZMO electric ‘oscillating’ mopping attachment. This is an active mop (uses the same disposable pads) that does a slightly deeper clean on hard surfaces. This is part of the JB Hi-Fi retail package – a more compelling reason to buy.
The box includes two sets of side brushes (whiskers) and a spare dust filter.
Setup – PASS
First, charge the robot fully (6.5 hours). Then download the Ecovacs app for Android or iOS. Set up an account and connect to 2.4Ghz Wi-Fi.
The app has the ‘usual’ overly legal, privacy policy and user agreement (date 06/08/2020). We tried to access these from the Australian website (because these are too long and tedious to read on a smartphone), but broken links prevent this.
You have no option but to accept these if you want to use the app and access the robots advanced features. If not, you can only manually activate the DEEBOT as a dumbot. I only point this out as users must read such terms and understand that your device collects location, video images, home layout, IP information and very much more.
If you later disable the AIVI smart detection function or remove location or camera access, then it reverts to a dumbot capable only of bump detection and pattern cleaning.
A note on privacy – this applies to most IoT device makers – FAIL
First, I am not having a go at Ecovacs per se – most IoT device makers have equally complex terms.
The purchase of an IoT device is to do a specific task – not as a massive data harvester or direct marketing facilitator. Ecovacs could set some standards here!
Information collected should be strictly limited to that which is necessary for the operation of the device. Optionally you should be able to reject telemetry to diagnose faults or improve future firmware updates. It is simply not good enough to state that by using the product, you agree to the wide-ranging terms.
Its privacy policy allows it to do pretty well anything it wants with your data. It can market or advertise to you; it could impose use charges; view your images; and so much more. It boils down to do you trust the company, or are you willing to blindly take the risk to use the product?
My beef is that Ecovacs does not make this clear before purchase and worse still, it is not on the website. I recall that the ACCC tore strips of Microsoft and Apple’s EULA some years ago for the very same reasons.
How does it work?
TrueMapping is Ecovacs marketing name for a dToF (Direct Time of Flight) laser that rotates 300rpm and maps out your home to a distance of about 3 metres. You will also see it referred to as LIDAR in other brands. It produces a 2D map that is relatively accurate and does not use ceiling or floor ‘landmarks’ as some lesser robovacs do.
It builds a map on first use (and you can update the map or make an advanced one) by traversing the room in a reasonably logical pattern. AIVI processes images from an onboard camera to help identify floor surfaces, obstacles and recognise many of the things that can trip a robovac up. The camera can also do remote viewing, and it has a sentry mode – probably good for stalking a pet.
It accepts up to three maps – handy for multi-floors, and you can set no-go zones. We understand that you must move the base station to each floor to make a map. You can also set up no-go zones.
We know all about Preparation for cleaning (see below) and in our opinion, the test home is as robovac safe as it can be. On the first mapping run, it reported 101 obstacles (and there were none). It gave the message “We recommend that you tidy up your home …”
On the second run, it reported 80 obstacles. On the third, 69 and the fourth, 61.
These obstacles show as little icons like cable, shoes etc. – and yet there were none. To be fair, the number reduced after a firmware update. I think there is more work to do here.
Preparation for cleaning – generic robovac advice
You cannot expect any robot cleaner to just to work! As a guide
Tests – Ecovacs DEEBOT OZMO T8 AIVI
Battery – PASS with caveats
The 14.4V/5200mAh battery claim is up to 200 minutes use (hard surfaces) and a 6.5-hour recharge from empty. We have to say that on both counts this is a lot longer than expected.
We love the run-time – it equates to about 50m2 per hour. Naturally, this reduces for a mix of hard floor and carpet (as it ramps up suction on the latter), but you should get at least 100-150m2 from it – unlikely that it will reach 200 minutes.
We dislike the recharge time – it is way too long if a whole-of-home clean is more than about 150m2. And there is no battery percentage indicator in the app so you don’t know if you can set off again on a different floor.
The battery model S10-LI-144-5200 is user replaceable (when out of warranty) requiring the removal of the base plate and several screws (not tested). It is vitally important to have ready access to reasonable cost replacement batteries as we have found that after about two years most robovacs lose too much charge capacity.
We cannot find an Australian retailer of either the genuine or third-party replacements. eBay Australia lists several, probably non-genuine suppliers for about $100 shipped from China.
Noise levels – PASS
It ranges from 62dB to 68dB with the Oscillating mop attachment. It is slightly noisier than other premium robovacs.
Dust bin – PASS
Homes have varying levels of cleanliness, so we vacuum weekly with an LG A9 Master or a Dyson V11 Outsize, and a Shark Ion for spot cleans to prepare for the robovac tests. We know roughly 400ml of detritus is what to expect.
In testing, we substitute that regimen with the robovac. This bin is small at 420ml is small filling after about 50m2 on mainly timber and ceramic tile floors. There is no bin full sensor/indicator, so when it is full, the device simply keeps going. Neato has a 700ml dustbin but no full sensor either.
Cleaning speed – PASS
It is a round device meaning that the main rotating brush is 16cm narrow. The dual rotating whiskers extend its reach to about 40cm.
By comparison, D-shape of square robovacs have a much wider brush (typically around 30cm) and edge whiskers extend that reach.
This robovac cleans about 1m2 per minute (similar to Roomba i7+) and the Neato D7 is 50% faster at 1.5m2 per hour.
It has 1500pa suction power. The Roomba i7+ is 1700, LG R9 and Neato D7 are 2000pa.
Cleaning depth/effectiveness – PASS
We set up test items (from fine to large) including lint, fine hair, muesli), 6mm square pieces of newspaper, frozen peas and Special K cereal.
It is fine with smaller items but rejects larger ones. It had a lot unsuccessful of attempts at frozen peas and Special K where other robovacs were fine.
The barefoot test (vacuum only) is good – no visible dust, but it did randomly miss smaller areas.
We then ran over the area with an LG A9 Master and found nearly 200ml more detritus. A test a week later with a Dyson V100 Outsize gave similar results. Most of that detritus comes from edges and the carpet areas.
Its really good on hard floors, not so good on edges and average on sisal carpet.
Cleaning under furniture – PASS
It will clean under furniture with approx. 100mm clearance.
Cleaning shag pile – FAIL
It will clean surfaces with up to a 15mm pile
Sill negotiation – PASS with caveats
It will negotiate door sills up to 20mm. It is relatively effective in backing off a sill if it is too high.
But if you have the multi-use pad installed it can get stuck on 20mm sills.
Wi-Fi – PASS
It uses a Wi-Fi N 2.4Ghz and is effective to about 30 metres from a home Wi-Fi router, less if that signal is through walls or you have a weak router. It did not drop out during the test. If you experience issues, it is more likely your router or Wi-Fi extender.
Base Negotiation – PASS
In all bar one case, it returned to base. The exception was using the oscillating mop where it will not return unless you remove it.
We could not test recharge and resume but assume it works.
Stair-drop – PASS
It missed robovac leap although it came precariously close as the sensors are under the front bumper. An extra sensor at the rear may prevent mishaps.
Error handling – PASS
On the whole, it is error-free. On one occasion it became stuck under furniture and could not find a way out (but it got in there). It sent a notice to my smartphone.
Edge cleaning – PASS-
We have seen better edge cleaning, especially under cupboard overhangs. It seems to place about a 20mm barrier between itself and the edge.
Mop – well two mops – PASS and PASS+
It comes with a standard passive mop plate that fits on to a removable water reservoir in the back. This can use either type of pad. It is more a wet pad that moves over the floor to collect dust. It does not remove stains or built-up grime.
More interesting is an oscillating mop unit (8 times per second) that is for use with the disposable pad only. The oscillating mop gives a better finish but still will not remove things like dried milk.
Both hold approx. 240ml of water that should be good for 100m2 of mopping. Water quantity is adjustable. There are no cautions against using a cleaning solution mixed with water.
The DEEBOT OZMO will avoid carpeted areas. There is one catch 22 here. If a tiled area is off a carpeted area (bathroom off a bedroom), you will need to move the robovac there and close the door.
In our experience, the oscillating mop is quite good for weekly use. The passive mop can leave water streaks.
Ecovacs claims the mop can remove 99.26% of bacteria (staphylococcus, bacillus subtilis and Coli bacillus) from the floor. Frankly, that is impossible with plain, room temperature water. It would require a strong 70%+ alcohol-based disinfectant in the water reservoir.
Filter – PASS
There is a three-layer filter on the dustbin. The claim is that it catches 99.4% of particle matter to 6um. But it is vital to properly clean and solidly tap the filer to get the dust out of the disposable (not washable) filter.
Maintenance – PASS
At this stage, there are no maintenance kits for the T8 on the Australian website. The US site lists the T8 as compatible with the 920/950/T5.
That kit (DX5G-KTA) includes a replacement rotating brush, two sets of whiskers and three dustbin filters for US $49.99. eBay Australia has it for A$99 including local postage.
The washable, reusable pads come in a pack of three and should last dozens of uses. eBay Australia has them for A$79.90 including Australian postage (way too expensive for microfibre fitted cloths). The only price for the disposable one-use pads is ex-China at $17.15 for a ten pack.
Warning – the majority of DEEBOT parts on Amazon, eBay, Kogan, Dick Smith and now Matt Blatt are not genuine and may affect warranty.
The maintenance costs are fair. Our test is not long enough to judge durability, but it seems very well made.
Voice assistance PASS but basic
It works with Google Assistant (tested) or Alexa via an installable skill.
You can ask it (give it a name) to
There may be other commands, but Google did not recognise the usual ones. Note that using a voice assistant exposes private data to those companies.
Camera – PASS-
We gave up on the camera. Obviously, DEEBOT needs to be out of the charging cradle for remote viewing. We have excellent internet speeds and strong whole-of-home coverage but kept getting time-outs. And frankly who wants to see feet.
Smart Butler mode lets you drive it around like a remote control car. There is a small scope here for home security and pet monitoring.
Known issues
Overall, it gets a good wrap from US users with the main issues being losing the map and losing Wi-Fi connection (probably the routers fault). Early users suffered from ‘double’ maps, and recent firmware updates fixed that.
The user reviews parallel our experience – a great robovac but average as a mop. That is probably as good as it gets.
GadgetGuy’s take – Ecovacs DEEBOT OZMO T8 AIVI is good but not perfect
On the plus side, it is amazing what a year of new tech can make. Mapping is fast and accurate. Obstacle avoidance is better, but there are still way too many false positives. It makes me question the need for a camera.
Otherwise, it is a typical robovac and a fairly slow, narrow brush one at that. Roomba i7+, LG R9 Master and Neato D7 blitz it as a vacuum cleaner. Being subjective – as a vacuum, this is a 7, iRobot an 8, LG a 9 and Neto a 10.
But the combo mop is interesting and almost unique. I guess it is a matter of if you hate vacuuming or mopping more. I vacuum weekly and mop monthly. But don’t expect the same results as a hand mop. The passive mop is akin to what the yanks call a ‘Swiffer’. The active mop is the reason to buy.
Overall, it’s a fine unit and one that almost passes every test criteria. Forget my privacy bent, but that is the biggest issue facing humanity (after COVID). Big Tech is going to come tumbling down when we have ultimate control of our data.
If you don’t need the camera/obstacle detection, the DEEBOT OZMO 950 is essentially the same power (1500pa) or the DEEBOT OZMO 920 is 1200pa with shorter run time. See a comparison chart here.
Rating the DEEBOT OZMO T8 AIVI
There are better straight robovacs that will vacuum faster and deeper especially if you have lots of carpet. Not hugely ‘must buy’ robovacs but we would be remiss not to mention that.
But as a robovac/mop combo it scores well in its uniqueness. But the caveat here is this observation is due to the oscillating mop version.
