Modern robotic cleaning systems, made under the Xiaomi brand and their sub-brands like Roborock or Dreame, have impressive navigation capabilities. However, out of the box, the device often operates in chaotic or sequential mode, without dividing the space of the apartment into logical zones. For the owner, this can mean that the vacuum cleaner will get stuck under the sofa in the living room while the bedroom is still dusty, or will endlessly ply through the corridor. To get the gadget to clean exactly where it is needed, and in a certain sequence, you need to conduct a map construction procedure.
The learning process of the device begins long before you first see a clear division into rooms in the app. You need to create the perfect environment for scanning space. It's not just turn on and forget, it's a full-fledged technical procedure that requires you to engage initially. The accuracy of the map depends on how well the room is prepared before the mapping algorithms are launched. If you ignore the training, the robot can combine the kitchen and the living room into one huge area or, conversely, break one room into several non-existent parts.
In this article, we will take a look at the whole journey from unpacking to creating complex cleaning scenarios. You will learn how lidars and gyroscopes work, why it is important not to interrupt the first launch, and how to use virtual restrictions. We will look at the nuances of the Mi Home application and alternative platforms such as Xiaomi Home or Roborock. Understanding the logic of the navigation system will allow you to not just poke buttons, but competently manage the process, saving time and battery life.
Preparation of the room for the first launch and scanning
Before a robot vacuum cleaner can effectively clean rooms, it must see their boundaries, using a laser rangefinder on the top of the housing, or a visual navigation system. To be accurate, it must clear the perimeter of the room. Remove wires, toys, socks and other small objects that may become entangled in brushes or be mistaken for obstacles. The base station must stand near the free wall so that the robot can travel far enough away from it for scanning.
Special attention should be paid to lighting and obstacle heights. If your model uses visual navigation (camera), it will not work properly in the dark. Laser models are less dependent on light, but shiny surfaces, floor mirrors and black carpets can knock down distance sensors. Black surfaces are often perceived by sensors as a cliff (ladder), and the robot will try to circumvent them, creating “dead zones” on the map where they should not be.
💡
Before the first run, be sure to wipe the sensors and lidar with a dry wipe - dust on the sensors can distort the map and lead to navigation errors.
It's important to have access to all the rooms. Open all the interior doors if you're planning to build a single map of the apartment. If the doors have high thresholds, make sure your robot's clearance allows you to overcome them. The standard height of the obstacle being overcome is about 2 cm. If the threshold is higher, the robot will consider the wall to be insurmountable and will not include this area in the map, which will make it impossible to clean the rooms in the future.
Map construction procedure and navigation calibration
After the preparatory work, the most important step is the construction of the map. In the Mi Home or Roborock application, you need to choose the mapping mode or simply start a full cleaning, before turning off the wet cleaning (if it is a combine harvester), so as not to wet the carpets during long maneuvers, the robot will start moving from the base, describing circles and returning to the start point to adjust the data. At this point, it is strictly forbidden to move the base, move furniture or take the robot on your hands.
⚠️ Attention: If you take a robot vacuum cleaner in your hands while you're building a map, it will lose orientation in space, it will break the map or create duplicate rooms. If the device is stuck, you better interrupt the task and resume it later.
The process can take anywhere from 30 minutes to 2 hours depending on the area of the room. The robot will repeatedly return to the base to recharge and continue scanning from the same point. This is normal behavior for SLAM algorithms (simultaneous localization and mapping). During this process, the device remembers the location of walls, furniture and stationary objects. Once completed, you will see in the application a black and white or color plan of your housing.
If the map is curved (rooms are skewed, walls float), you can reset the procedure. To do this, select the option "Reset the card" or "Delete the card" in the card settings. Then repeat the scanning procedure, making sure that nothing has changed in the room. Sometimes it helps to start at night, when the house is quieter and no one walks, disturbing the sensors.
Dividing the map into rooms and zoning
When the map is done, it often happens that the robot sees the entire space as one big room. To teach it to clean rooms, you have to manually divide the space into zones. In the application interface, find the "Maps Edit" or "Room Management" button. There's a "Separate" tool available here. You have to draw a line through the doorway, separating, for example, the kitchen from the hallway.
The separation line must run exactly along the walls or through the center of the doorway. The virtual walls created when split will be invisible to the robot, but it will perceive them as a solid obstacle. This allows you to set different cleaning modes for different areas. For example, in the kitchen you can set maximum suction power, and in the bedroom you can set a quiet mode. After separation, each zone is assigned a name and type of room (kitchen, bedroom, living room).
☑️ Zoning checks
If automatic room recognition fails, you can use the Merger tool to merge the wrongly separated parts of a room, and you can also use the Rename feature, which is important for voice control. If you tell a smart speaker to "clean up in the kitchen," the robot needs to figure out which zone on the map matches that name. The accuracy of naming is critical for smart home scenarios.
Configuring virtual walls and forbidden areas
Beyond the physical separation of rooms, there's the concept of virtual constraints, which are necessary to protect certain objects that the robot should not touch, and in the app, it's called Virtual Walls or No-Go Zones, and you can draw a rectangle around a long-pile carpet so that the robot doesn't drag it into itself, or you can designate a zone around a pet bowl.
There are two main types of restrictions:
- 🚫 No-go zone: The robot will not enter this perimeter under any circumstances, and is used to protect areas where cleaning is unnecessary or dangerous.
- 🧱 No-mop zone: The robot can drive through this area, but it won't include wet cleaning mode. Perfect for carpets if you don't want to remove them before washing the floors.
You can adjust these areas in map editing mode, you can choose the type of restriction, you can draw a rectangle of the right size with your finger, and you can save the changes. And the important thing is that these constraints are tied to the coordinates of the map. If you move the sofa that was the guide for the virtual wall, the robot will still be going around the old place, so virtual walls are better tied to stationary objects, such as table legs, columns or corners of the walls, rather than moving furniture.
What if a robot ignores virtual walls?
Cleaning scripts and room schedules
Once the map is divided and the zones are set up, you have flexible control, you're no longer limited to the Start button, you have the Select Room or Area Clean menu, and you can point your finger at a particular room on the screen, and the robot will go to just clean it, saving battery life and time.
To automate the process, you use Schedule, and you can create multiple profiles.
- 📅 Daily cleaning: Every day at 10:00 a.m., a robot cleans the kitchen and corridor.
- 🛋️ Deep cleaning: On weekends at 12:00, the robot cleans the entire apartment with maximum power.
- 🌙 Night mode: Quiet bedroom cleaning only when you are not at home (if integrated with geolocation).
In advanced settings, you can set the cleaning sequence, for example, the kitchen (where it is always dirtier), then the corridor, and at the end of the living room, and you can adjust the suction force and the amount of water you can supply to each room separately, and this is especially true for models with electronic water supply, where it is important not to wet the laminate in the bedroom as much as the tile in the bathroom.
Comparison of models and navigation capabilities
Not all Xiaomi robot vacuum cleaners are equally good at mapping, hardware differences determine zoning accuracy and speed, and a chart below compares the main types of navigation found in ecosystem devices.
| Type of navigation | Map accuracy | Working in the dark | Examples of series |
|---|---|---|---|
| Laser (LDS/Lidar) | Tall. | Yes. | Roborock S5, Xiaomi Vacuum Mop 2 Pro |
| Visual (vSLAM) | Medium/High | No (light needed) | Dreame Bot L10s, Xiaomi Vacuum-Mop 2 Lite |
| Gyroscopic | Low (no map) | Yes. | Xiaomi Robot Vacuum Essential |
| Chaotic. | Absent. | Yes. | Older Mi Robot 1S models (without Lidar) |
The gyro models don't map the space as they know it. They move the snake algorithm, but they don't remember the layout. You can't teach them to clean rooms, they either clean the entire area at once or they work on a timer. You need a model with a laser rangefinder or a high-quality camera to fully zoning.
💡
Only models with a laser (LDS) or advanced visual (vSLAM) sensor can save the map and divide it into spot cleaning rooms.
Problem Solving and Map Maintenance
In the process, the map can go, which happens if you move the base to another place without updating the tether, or if the robot has been towed on the carpet for a long time. Symptoms: the robot thinks that it is in the living room, although it is physically in the bedroom, in which case the function "Update Position" (if available) or a complete reset of the map and rebuild helps.
Clean the sensors regularly. Dust on the lidar window or on the bottom panel of the drop sensors leads to errors. Also check the wheels - the coiled hair can move the robot to the side, which will knock the navigation algorithms. If the robot began to leave uncooked stripes in the center of the room, perhaps the problem is not the map, but the wear of the main brush.
⚠️ Warning: Do not wash sensors with a wet cloth immediately after cleaning if they are heated. A sharp temperature drop and water entering the housing can cause electronics to fail. Use dry wipes or slightly damp fabric for external wipes.
To keep the map up to date, it is recommended to run a full cleaning of the entire apartment at least once a month, which allows the robot to check the current state of affairs with the saved plan and adjust routes if you rearrange a chair or add a new nightstand. Modern Xiaomi algorithms are able to recognize changes in the environment, but they need regular practice.