Xiaomi’s robot vacuum cleaners and their sub-brands like Roborock or Viomi are famous for their sophisticated navigation system. But even the most sophisticated map-building algorithm is powerless against the dust layers that accumulate over time on optical sensors. If your smart assistant starts to stumble across obstacles, get stuck in a flat place or randomly rush around the room, the problem is often not software failure, but simple pollution.
Regular maintenance of optical elements is key to the long-term service of the device. Dust, pet hair and small debris can completely block the access of light to photodiodes, which makes the system perceive the clean floor as a cliff or wall. In this article, we will discuss in detail where the sensitive elements are, how to wipe them and how to avoid typical maintenance errors.
It's important to understand that distance sensors and elevation gradient sensors (cliff sensors) work on the principle of infrared radiation. Any film of dirt distorts the signal, making the robot think that there is a wall ahead, which is not, or, conversely, do not notice the step of the ladder. Proper care of them does not require complex equipment, but requires accuracy and understanding the physics of the process.
Types of sensors and their location in Xiaomi models
Before cleaning, you need to localize the problem areas. Xiaomi devices use a combination of different sensors, each of which is responsible for its own area of work. The upper "turret" is a lidar (LDS), which rotates and scans the room with a laser. The side and bottom holes hide the IR sensors of collisions, and the "abdomen" are floor sensors.
Visual navigation (VSLAM) models have a camera instead of a lidar that also requires cleanliness. Lens contamination results in a complete loss of orientation in space. At the same time, classic gyroscope models rely solely on side sensors and wheel data, so cleaning them is critical for straightness.
Notice the bottom of the hull. That's where the cluster is. 3-6 And these are holes that have cliff sensors hiding inside, and they prevent the device from falling down the stairs, and if they're clogged with dust, the robot will think it's above the cliff, and it will refuse to go forward, making a mistake or just turning around.
- 🔴 lidar (LDS) — laser rangefinder on the top panel for map construction.
- 🟡 IR sensors bumper – respond to the approach to furniture and walls.
- 🟢 Cliff sensors – lower altitude sensors for safety.
- 🔵 Wheel sensors are optical encoders that count the distance traveled.
Visually, it's easy to distinguish a dirty sensor: if you see a gray plaque or villi inside the hole, then the signal is blocked. The transparent window should shine clean. Remember that in some models, the sensors are protected by a thin film that also needs to be wiped, but in no case should not be removed.
Necessary Tools for Safe Cleaning
You don't have to be an electronics engineer to maintain your sensors, but you have to have the right toolkit. Using the wrong materials can cause micro-scratches on the lenses that can then scatter light and make navigation worse. The main goal is to remove dust without damaging the sensitive surface.
Electronics' main enemy is moisture and static. So microfiber is the best choice for wiping. Conventional paper wipes can leave a pile that will clog the sensor again, and rough fabrics can scratch plastic. Alcohol-containing liquids are permissible, but in minimal amounts and only to remove greasy stains if the robot cleans the kitchen.
For hard-to-reach places where a cloth can't get through, compressed air in cans or a pear for cleaning optics is ideal. A powerful jet of air knocks compressed dust out of the depths of the sensor well. You may also need a cotton swab, but you need to use it with extreme caution to keep the villi from tearing off.
☑️ Cleaning tools
It is strictly forbidden to use a vacuum cleaner to clean the sensors themselves, a powerful flow of air can damage the membranes of the sensors or introduce static electricity that will disable the electronics, and mechanical impact should be minimal.
⚠️ Warning: Never pour liquid directly onto the robot body or into sensor holes. Only a napkin lightly moistened with water or special lotion for screens should be wet.
Step-by-step: cleaning lidar and side sensors
So, to start with the most visible element, lidar, this is a rotating turret on the top cover, and over time, a layer of dust settles on the laser's protective glass, which reduces the scanning range, and first, gently wipe the top cover with a dry microfiber in circular motions.
Then, look at the side walls of the lidar module itself, which often accumulates dust in the cracks, use a dry soft brush to remove the dirt from the moving parts, and if the turret is removed (in some Xiaomi Vacuum Mop models), you can gently blow it with air, but it is better not to disassemble the device without having to knock the calibration.
The side IR sensors are located around the perimeter of the bumper. These are black or dark gray windows. Wipe them with a wet (not wet!) wipe to remove the stains. If there are sticky traces of sweet soda or syrup on the surface, you can use a drop of the window washing agent on the rag, but then be sure to wipe dry.
What do you do if lidar is noisy?
Checking the quality of the cleaning is simple: turn on the robot and start building the map. If the lines on the map are flat, and the robot does not "scramble" from invisible obstacles, then you did great. Otherwise, repeat the procedure by paying more attention to the side edges.
Maintenance of lower sensors and wheels
The bottom of the robot vacuum cleaner is a high-risk area, and here the cliff sensors are constantly exposed to dust from the main brush, flip the device over and find a number of holes on the bottom, and inside each one there is an infrared emitter and receiver.
Use a cotton swab to clean, and carefully, without pressing too hard, wipe the inside of each hole, and you'll be surprised at how much gray dust there is, and even the thinnest layer of plaque can block the IR signal, making the robot think it's hanging over the stairs.
Don't forget the sensors on the wheels. They're in the recesses of the wheel niches. If they're contaminated, the robot can miscalculate the distance traveled, causing map errors and missing entire rooms when cleaning. Scroll through the wheels and wipe the visible parts of the sensors.
| Type of sensor | Location. | Symptoms of pollution | Tool. |
|---|---|---|---|
| Lidar (LDS) | Upper tower | A torn card, loss of position | Microfiber, brush |
| Cliff sensors | Lower panel (closer to the edge) | Refusal to go forward, fall error | Cotton swab |
| Bumper sensors | Frequent false collisions | Wet wipe. | |
| Wheel optics | Inside the wheel niches | Chaotic movement, map errors | Dry stick |
After cleaning all the lower elements, be sure to wipe the charging contacts on the bottom.Oxidation or dust on the contacts can cause the robot to not charge or lose contact with the base.
💡
Wipe the bottom sensors every time you empty the dust container, which will take 10 seconds, but save you 90% of navigation problems.
Cleaning the camera in models with VSLAM
Owners of visually navigable models (such as the Xiaomi Vacuum Mop 2 Pro+ or Roborock S7 MaxV) should pay special attention to the camera. The camera is usually located on the top panel or on the turret. Any scratch or greasy spot on the lens turns a smart robot into a blind kitten.
To clean the lens, use only a special fabric for optics or high-quality microfiber. Movement should be from center to edge, without strong pressure. Do not use alcohol in its pure form, as it can damage the anti-reflective coating of the lens.
If the robot is getting lost a lot or the map is being distorted, the first thing to do is to check the camera's cleanliness. Sometimes the problem is solved by simply wiping it out. Also, check if the lens is covered with a protective film that you forgot to remove when you bought it, which is a common mistake for beginners.
⚠️ Warning: If after cleaning the camera, the robot still emits audible signals about a navigation malfunction, there may be a problem in the software or hardware damage to the camera plume.
Frequent Mistakes and What Not to Do
The desire to help the device sometimes backfires. The most common mistake is using aggressive chemistry. Acetone, solvents, or ammonia window cleaners can make the plastic cloudy and brittle. The cloudy plastic scatters the IR rays, and the sensor stops working properly.
Another mistake is ignoring humidity, and if you wipe your sensors with a wet wipe and immediately turn on the robot, moisture can get in and cause a short circuit. Always give the device 5-10 minutes to dry after wet cleaning the sensors.
Don't try to pick inside sensor holes with sharp objects like needles, pins or toothpicks. You can damage the emitter itself or scratch the inner surface, which will create permanent glare for the sensor. Use only soft materials.
- ❌ Don’t use paper towels – they leave a pile.
- ❌ Do not blow with your mouth - saliva will fall on the lens and dry with a crust.
- ❌ Do not disassemble the lidar module without experience - calibration will go wrong.
Remember, regularity is more important than intensity, so it's better to wipe your sensors with a slightly damp cloth once a week than to try to clean up petrified dirt once every six months with aggressive methods.
💡
Use only dry or slightly moist microfiber, and aggressive chemistry and sharp objects are the main enemies of your robot's navigation.
Diagnosing problems after cleaning
How do you know if the cleaning helped? Put the robot in map-building mode, or just send it to the cleanup. Notice the behavior: if it stopped being dumb at the couch, doesn't spin around and confidently drives to the base, then the procedure was successful. In the Mi Home or Roborock app, you can see the history of the cleaning - the map should be flat, without porridge from the tracks.
If the problems persisted, maybe the pollution wasn't the only factor, check if the garbage got stuck in the wheels, if the hair was wound onto the side brush, and sometimes the sensor is functional, but it's closed with some foreign object, like a sticker or a piece of insulation.
In rare cases, the sensor may fail physically, and if after careful cleaning and rebooting, the robot continues to produce sensor errors (such as a wall sensor or a drop sensor), it may need to replace the module, but in 95% of cases, it helps to clean the optics.