Today, we’re doing a head-to-head vacuum challenge between the iRobot Roomba 980 and the Xiaomi Mi robot vacuum. Xiaomi introduced its first robot vacuum to market not too long ago and with it comes a brand new AI based navigational routine and a design that certainly looks familiar when comparing it to others out there. Does the reduced price of Xiaomi’s effort mean reduced cleaning ability? Let’s put it against the industry veteran and see which one comes out on top.
Our first test covers the spot cleaning modes for each vacuum and their ability to clean up small messes that can happen around the house. Xiaomi touts the Mi robot vacuum as having actual AI instead of just a good cleaning algorithm and it’s interesting to watch the vacuum learn as it goes.
The spot cleaning mode is a bit hidden over some other vacuums and is enabled by a long press of the home button on the top of the vacuum. From here, the vacuum appears to work out an approximately 5 by 5 foot square around an area that it’s placed in. It works its way around the table legs and continues on to the end of the virtual square and completes the entire spot without fail. There were a few oddities in the beginning where it actually missed a spot. But then it went back later and finished that part before the end of the cleaning cycle.
Moving on to the Roomba. The Roomba spot-clean mode is actually called a target clean mode and for good reason. The vacuum literally moves in a circular target position around a center point where it starts, presumably cleaning up small spills in a very specific area. iRobot seems to have tweaked this mode a little bit since our last vacuum tests but it still has the same faults in that it really cannot work around objects or obstacles in the way.
Likely, if there’s a spill someone would probably clear the area as much as possible before sicking a vacuum on the mess. But this is still a fault that could be remedied by better navigational aspects in this particular mode. A clear win for Xiaomi in this particular test.
The next test puts the navigational abilities of each vacuum to the test to see if they can get through a room with lots of obstacles and small places to go. Starting with Xiaomi, the navigation for the Mi robot vacuum begins by scanning the room first, followed by a perimeter sweep all the way around and into every crevice the vacuum into. Notice how close it comes to the sides of the walls and objects without really rubbing against them.
After it sweeps the perimeter, it moves inward in rows keeping from overlapping the sides that it just swept. In this scenario, the construction toy here on the right seems to have folded a little bit as it repeats a section on the cars rug. However, it gets quite interesting when the vacuum reaches the leg of the crib here as it makes a full sweep around it to ensure proper cleaning instead of including this in its sweeping rows. This is evidenced by the vacuum completely avoiding the circular path around the leg in the next row of steps.
Once it reaches the ottoman in the middle, it attempts to go around it and resume the sweeping rows. However, it stumbles a bit on the darts laying out on the floor on the left side but doesn’t actually pick them up. Something very strange happens. Once it makes its way back around to the tiger, though, the navigation almost seems to be a bit confused as it rumbles around seemingly aimlessly and doesn’t really finish sweeping its rows. Rather it goes back to exploring the perimeter room again as if something has changed drastically.
In its second pass around the room, the vacuum actually manages to suck up two of the three darts on the floor and successfully push the other toys out of the way to cover the entire floor on the process. Sucking up the darts, though, made a rather nasty sound and it’s entirely possible that this could significantly impact the vacuum’s performance if this were to happen in the beginning of a cleaning cycle.
Moving on to the Roomba, iRobot’s navigation is very different from Xiaomi’s and you’ll notice this by the fact that it not only starts immediately without scanning the room first but that it also begins in rows instead of making a perimeter. The logic here lies in the completion of a straight row and any obstacles in the way of the row are sometimes worked around and sometimes used as a way to end the row.
Initially, the Roomba gets under the crib but there’s a line straight behind the leg that it never even touches because it ended the row prematurely. When it reaches the leftmost side of the room, it tries to finish out the perimeter of the area, and then comes back around for some seemingly random places it believes it missed in the mapping process.
After this piece is finished, it goes back and completes the perimeter last, sweeping anything up that might have been missed while working on the inside first. The second pass around the room moves those rows into columns and it does a better job of logically covering the area than it did in the first time. In the entire test, there’s only that one place that it doesn’t seem to get any cleaning at all. It’s that same spot we talked about earlier behind the crib leg that it missed that first time around.
The Roomba also only sucked up a single dart and left the rest at the perimeter behind the chair. This test shows just how much better Xiaomi’s navigation can be at times, grabbing every nook and cranny there is but the design of iRobot’s rollers and suction kept the Roomba 980 from getting hung up on the small darts and making lots of awful grinding noises in the process.
Still, while both vacuums picked up the darts, Xiaomi’s Mi vacuum eventually got them all the way into the storage bin while the Roomba’s got lodged in the path to the bin likely causing suction issues far more than a full storage bin would. This one’s a win for Xiaomi.
Our third test is the dirt on carpet test, a test that may not be the easiest for a vacuum that relies on a battery-powered motor. The carpet was vacuumed before the test and in between each vacuum is run. We use the rainbow with a carpet attachment to it for extra cleaning. We also made sure to clean out each vacuum’s filters, rollers, and systems to ensure the highest performance for each one showing how we perform in the real world since these vacuums have been used regularly in the home but have been cleaned as you normally would if you owned one.
As with the previous time we were in the test, we sifted the dirt first to ensure that what we’re putting in the carpet is nothing but small particles. There’s no sticks or other debris to be found. Half a cup of this was sifted on the carpet and an approximately 5 by 5 foot area was marked off for the test.
Starting with the Roomba, as we saw in the previous test, the Roomba starts right up and goes, covering the middle of the area it’s placed in until it hits a wall or obstacle in the way. What’s unique about the Roomba is the dirt sensor placed underneath the unit which is designed to detect when large amounts of dirt or debris are making their way under the vacuum. And on the third row, you can actually see this dirt sensor kick in. The Roomba 980 backs up and rolls over the large dirt spot three times until it’s satisfied that enough stuff has been picked up.
It then moves around the perimeter for sweeping and finishes with a column of sweeps through the middle of the floor, again, to pick up nearly all the dirt that was dumped on the carpet. This is consistent with the previous test we did versus the Neato botvac connected and also proves that the Roomba is consistent in its cleaning results. It’s also worth noting that the Roomba 980 in particular has the ability to adjust its suction power based on the type of surface it detects. So in this case, it was running at its highest and loudest suction because it’s on carpet.
Moving on to Xiaomi. Per usual, Xiaomi’s vacuum scans the room and moves around the perimeter first defining the area it will be cleaning in followed by a series of rows and columns to get the floor extra clean. To make things even here, we made sure to select the maximum suction setting on the Xiaomi’s Mi home app for the Mi robot vacuum, and on the first pass, it actually looks as though the Mi robot vacuum would lose quite handily to the Roomba 980.
This is a bit surprising given that Xiaomi actually has bristles on its roller rather than a completely silicone ones iRobot uses. But the second pass around the room evens things out a bit here. In the end, it’s almost neck and neck between the two vacuums. However, the Roomba 980 does manage to extract a little bit more from that carpet, giving it a slight win in this category.
By popular demand, we’re throwing out a few pet hair tests here. This first one’s going to be on carpet. For this test, we filled a cup with freshly brushed hair from a golden retriever, something that’s famous for sticking to just about everything in sight. We broke this clump of hair into six pieces and rubbed them into the carpet using the same space we used for the previous dirt test. Remember these carpet and vacuums are thoroughly cleaned between the tests to ensure the most accurate end results.
Watching the Roomba work its way through the first two rows of cleaning results in a rather interesting problem that crops up. The initial sweep of the hair makes it look as though the Roomba sucks everything up immediately. However, after the second clump is picked up, we see a smaller piece apparently fall off resulting in a smaller clump that gets left behind. Whether or not this was caused by the rollers having difficulty holding on to the hair or not is certainly a great question.
And even finishing out the round here, we see this single clump that was left behind is the only piece left on the carpet by the Roomba. It’s definitely able to grab most of the hair but this does make one wonder whether or not this would happen in more situations.
Moving on to the Mi robot vacuum. Xiaomi’s differently designed rollers actually contain a brush rolled around a silicon band giving it a dual action purpose that’s nearly identical to the one Neato includes on its botvac line. It appears that, even though there’s only one roller on the Xiaomi instead of the two rollers as the Roomba has, the Xiaomi does a better job of grabbing every piece of hair on the first pass and not actually letting go of it throughout the test.
In normal circumstances, the vacuum will only make a single pass of a room unless set otherwise so this is the case of where you might want to enable two pass cleaning if you’ve got pets. Otherwise, it seems the Xiaomi Mi robot vacuum takes a slight upper hand here in grabbing hair that first time around.
Hardwood and tile floors are certainly more common nowadays than carpet would be so this next one tests hair on a hard surface, which is going to be difficult for a vacuum because there’s no friction to keep that hair in place like carpet has. We took a pile of hair and split it evenly into two large chunks, spreading the chunks over a small tile area for each vacuum to clean up.
The initial perimeter sweep from Xiaomi’s vacuum helps clean out any hair on the corners and edges of the walls before the vacuum heads into the center of the room for interior clean up. The vacuum cleans up all but a few scant hairs on the floor as well as a small chunk that was left behind in the interior cleanup portion of the test. As we saw with the previous carpet test, Xiaomi’s robot vacuum really does an excellent job in general with hair, but the errant hairs left on the floor here and there definitely give us cause for concern if this is used in a home with pets that primarily has hard surface floors.
As usual, iRobot’s Roomba 980 works in rows first, sweeping the interior and moving to the perimeter afterward to try to grab those small pieces that it might have left on its way in the initial run. While making the turns, we can see the vacuum side brush does an impeccable job of grabbing the large chunks of hair and shooting them inward towards the Roomba’s dual sweeping brushes in the center.
What’s really interesting right here is that we see the Roomba shoot out one of the chunks of hair that seems to have gotten stuck on the side brush. And unfortunately, in this case, it got shot under the couch where the vacuum won’t be able to get it. As it moves through the rest of the room, however, it picks up every single other piece of hair without error and even grabs those smaller hair particles that Xiaomi’s Mi robot vacuum was not able to do.
At the end of the test, the only real piece of hair left was the chunk that got thrown under the couch. But the actual floor itself was cleaner since it didn’t have a single hair particle scattered around as Xiaomi’s vacuum left. It’s a little difficult to choose a winner here, although, since the majority of the floor was cleaner after the Roomba ran, we’re going to give Roomba the slight edge here.
This last test focuses on large debris, specifically on hardwood surfaces, like tile, where debris could be easily pushed around instead of being vacuumed up. Chex cereal is the debris of choice here as it proved in our last round of comparisons to make for a rather difficult thing to pick up. For this test, we used a half cup measurement of Chex and tossed it into the middle of a 5 by 5 foot tile area, which was blocked off for the test.
The first row made here shows what a great job that large side brush can do when it comes to moving debris right into the path of the Roomba sucking up every piece straight through. That second row also shows the debris sensor being tripped and we see a rather interesting behavior happen after the debris sensor is tripped yet again at the end of the row. After trying several times to suck up the Chex and failing, it actually lunges forward into what appears to be a last-ditch effort to quickly grab those pieces before they get away. It then spends quite a bit of time over these pieces until they are actually all picked up and then moves on to the rest of the floor.
Throughout the rest of the cleaning process, we see it consistently moving back and forth over the Chex pieces as it continues to attempt to clean them all up. But at the end of the round, it seems to have completely forgotten about the corner and the side near the door, even spitting a few pieces back out when it finishes up. Unfortunately, there are not only pieces left around the floor but there are also fragments of pieces hanging around the edges of the squared off area as well. This is definitely a different result than we saw with the dog hair test.
Moving on to Xiaomi. Per usual, the Mi robot vacuum and its AI powered logic begins by making a perimeter sweep followed by evenly laid out columns and rows to clean out the middle. Upon the first pass, the Mi robot vacuum actually cleans up the entirety of the mess minus a single Chex piece in the middle of the floor, and then promptly returns to pick it up. At more than a minute quicker than the Roomba 980, the Xiaomi Mi robot vacuum actually manages to pick up every single last morsel, including any and all pieces that might have gotten crunched in the process.
It’s pretty incredible to see just how efficient Xiaomi’s vacuum is and this test proves that it’s better with these kinds of debris on hard surfaces than iRobot’s vacuum is. In the end, the Xiaomi Mi robot vacuum has claimed four victories over the course of six tests and it’s clear that Xiaomi has engineered a pretty incredible vacuum here. Seemingly borrowing the best elements from iRobot, Neato, and some others, as well as throwing in plenty of their own unique design traits and a new AI based algorithm inside.
The Xiaomi Mi wins in many of the tests, although it’s probably worth considering the Roomba 980 over Xiaomi’s vacuums if you have pets and hardwood or tile floors. There’s more to these vacuums than just cleaning though, so check out our comparison article to see which vacuum is the easiest to clean and maintain, which filters are best, and which one’s behavior and attitude will work better in your home.
|Best of the Best!||Budget Pick||Great for Carpets|
|Roomba i7+||Eufy 11S (Slim)||Roborock S4|
| • 75min Runtime
• Keep Out Zones
• Clean Base
| • 100min Runtime
| • 150min Runtime
• Precision Navigation