Have you ever noticed an unusual car slowly cruising through the streets of your neighborhood with a bulky structure on its roof? Hidden inside this installation are powerful cameras, lidars and GPS modules that record the surrounding space with the highest accuracy. Such vehicles have become an integral part of modern urban infrastructure, collecting terabytes of data for digital maps.
Drivers often react to them in different ways: some wave their hand, some try to go around, and some simply ignore them, considering it another experiment. However, there is complex technology behind this process that directly affects how you build routes in the navigator and how security systems work. Geodata collection is the foundation on which smart cities of the future are built.
In this article, we'll look at who's behind these cars, what kind of information they collect, and why they sometimes seem to drive in the same circle. Understanding these processes will help you better navigate the digital world and understand how your data (and your car data) gets into global databases.
Who puts cars with cameras on the roads?
The main players in the panoramic image collection market are technology giants. The leader here, of course, is Google with its project Google Street View. Their fleet consists of specially converted vehicles, often in bright colors, that can be found almost anywhere in the world. They set the standard for quality and coverage of the territory.
However, do not forget about other competitors. Apple is also actively developing its service Apple Look Around, using their own cars with characteristic domes on the roof. Russian analogue - Yandex, whose cars with a characteristic yellow color and equipment regularly update maps of cities in the CIS countries. Each of these companies has the goal of creating the most detailed digital copy of the real world.
In addition, there are specialized mapping companies such as Here Technologies or TomTom, which provide data for car navigators and government agencies. Their cars may look less flashy, but they perform the same function - updating the road situation.
⚠️ Attention: Vehicles of data collection companies always have appropriate markings and contact numbers on board. If you see an unmarked car with suspicious equipment, you should be vigilant.
Sometimes third-party contractors or volunteers are involved in data collection, using special backpacks with cameras for pedestrian areas. This allows you to cover parks, shopping centers and narrow streets where a standard car cannot pass.
Technical equipment: what is on the roof?
The structure on the roof of a car is not just one big chamber. This is a complex engineering complex that includes many sensors. The basis is an array of several high-resolution cameras, directed in different directions to create a 360-degree panorama. Usually their number varies from 8 to 15 pieces depending on the generation of equipment.
Besides optics, a critical element is LiDAR (laser scanner). This device emits laser pulses and measures their return time, creating an accurate three-dimensional model of space. Lidar allows you to determine distances to objects, the height of curbs, signs, and even the dimensions of parked cars with centimeter accuracy.
To link all data to coordinates, a professional GPS/GLONASS antenna is used, often located in the center of the dome. The system can also include accelerometers and gyroscopes that track the vehicle’s position in space, which is especially important in tunnels or places with poor satellite signals.
How is data processed on the fly?
Typically, raw data is written to hard drives inside the vehicle. Pre-processing may occur to check the integrity of the recording, but full editing of panoramas and linking to the map is carried out in data centers after the end of the trip.
All equipment must be securely fastened and protected from vibration, dust and temperature changes. Inside the cabin there is also a computing unit and storage systems that require cooling, so the car can be noisier and warmer than in a regular car.
Why do you need to constantly update maps?
A city is a living organism that changes daily. Yesterday there was road repair here, today a new shopping center opened, and tomorrow the traffic pattern at the intersection will change. A static map is useless for a navigator, who must guide you according to current rules. That is why a car with cameras must regularly drive along the routes.
Data updating is necessary for the correct operation of artificial intelligence algorithms, especially in the context of the development of unmanned vehicles. Autonomous cars need highly accurate maps that indicate not only the presence of a lane, but also its markings, speed limit signs and even the condition of the road surface.
In addition to navigation, up-to-date data is needed for emergency response services, logistics companies and delivery services. An error in the map can cost time, fuel, or even lead to an accident if the truck is driven on a road with a height restriction that is not displayed in the system.
- 🚧 Road works: recording temporary and permanent changes in traffic organization.
- 🏢 New objects: the emergence of residential complexes, office centers and points of interest (POI).
- 🚦 Road signs: installing new speed limits, stop signs, or paid parking zones.
- 🛣️ Coating condition: detection of potholes, speed bumps and changes in the road profile.
The frequency of detours depends on the dynamics of changes in the area. In the center of a metropolis, a car with cameras may appear several times a year, while in residential areas or rural areas the intervals may be one to two years.
Process of collecting and processing information
Routes for cars are planned in advance by algorithms that analyze the age of the latest data and received user complaints. The vehicle operator (driver) follows a given track, trying to maintain a uniform speed and avoid sudden maneuvers that could blur the image.
After completing the shift, the data is transferred physically or through secure communication channels to cloud storage. This is where the automatic processing phase begins. Computer algorithms stitch together images from different cameras, level the horizon and correct colors.
☑️ Data processing stages
One of the most important stages is anonymization. Artificial intelligence automatically finds people's faces and car license plates to blur them before publication. This is a required procedure to comply with privacy laws in many countries.
However, a fully automatic process is not ideal. Sometimes the system misses a number or blurs part of the background. For such cases, mapping services have tools that allow users to request additional blur for a specific object or building.
| Parameter | Description | User value |
|---|---|---|
| Camera resolution | High (often 4K and higher per sector) | Ability to read street name or store sign |
| Refresh rate | From 6 months to 2 years | Up-to-date information about new roads or junctions |
| GPS Accuracy | Centimeter (with correction) | Precise positioning of the entrance to a building or parking lot |
| Shooting speed | 30-60 km/h (optimal) | Picture quality without blurring fast moving objects |
Impact on the development of unmanned technologies
The data collected by cars with cameras is the “fuel” for training the neural networks of self-driving cars. In order for a robot driver to drive safely around the city, it must “see” and “understand” the world around it in the same way as a human, but with greater accuracy.
The collected panoramas are used to create HD maps (high precision maps). Unlike conventional navigation maps, HD maps contain information about each element of the road infrastructure: type of markings, height of traffic lights, steepness of slopes. This allows the autonomous car to plan its trajectory in advance.
Without regular updates of such maps, using autopilot in difficult urban conditions would be impossible. The car simply would not know that the traffic pattern had changed ahead or that a temporary sign had appeared. Therefore, every kilometer traveled by a filming vehicle brings the era of mass robotaxis closer.
Data collection for maps and for drones are complementary processes, where the quality of the data directly affects the safety of future autonomous transport.
In addition, based on this data, various driving situations are simulated for virtual tests of control algorithms. Engineers can digitally recreate a real intersection and test how the drone will behave under certain conditions.
Privacy and legal aspects
Privacy remains one of the most debated issues in the context of street photography. The laws of different countries regulate the collection of images in public places differently. Europe has strict GDPR regulations that require minimizing personal data, while other regions may have more flexible rules.
Operating companies are forced to invest huge amounts of money in blurring technologies. Algorithms learn to recognize not only faces and license plates, but also, for example, open smartphone screens in the hands of pedestrians or numbers on houses, if the owners require it. Map users can often request that their entire home be blurred if they feel the filming is disturbing their peace.
There are also restrictions on filming sensitive facilities, military units and some government institutions. The routes of such vehicles are agreed upon in advance to avoid entering prohibited areas. Violation of these rules may result in serious legal consequences for the operator.
It is important to understand that filming is carried out exclusively from public areas. Cameramen do not have the right to look through a fence into a private yard or film through windows, and such footage is usually deleted during moderation or based on complaints.
How can regular drivers help?
While professional data collection requires expensive equipment, everyday users also contribute to keeping maps up to date. Many navigation apps allow you to report accidents, roadworks or new cameras in real time. This data is instantly displayed on other users' maps.
There are programs that allow you to upload geotagged photos to help cartographers verify changes. For example, if a store is marked as “closed” on the map, and the user uploads a photo of a working sign, the algorithm will prompt you to update the object’s status.
Drivers can also report routing errors or missing road signs through special feedback forms in the apps. Each edit is reviewed and, if approved, added to the database, making navigation better for everyone.
- 📸 Photo recording: Uploading current photos of building facades and entrances.
- 📝 Editing: correction of street names and house numbers.
- ⚠️ Signals: prompt notification of obstacles on the road.
- ✅ Confirmation: checking the existence of objects added by others.
Thus, the map ecosystem is built on a symbiosis of professional equipment and crowdsourcing. A camera-equipped machine sets an accurate geometric basis, and users fill it with live, up-to-date information.
Why can a car with cameras drive very slowly?
Low speeds (often 20-40 km/h) are necessary to obtain clear, blur-free images, especially in heavy city traffic. In addition, the equipment must have time to read data from lidars and GPS with a high sampling rate.
Is it possible to overtake a car with cameras?
Yes, you can, if the traffic situation and traffic rules allow it. However, it is worth remembering that sudden changes in lanes near the filming vehicle can ruin the panorama, as you will end up in a “blind spot” or be blurred by algorithms. It is better to overtake smoothly and at a safe distance.
What should I do if I accidentally got into the photo?
In most cases, your face and license plate number will be automatically blurred before publication. If you think that the blur is not of sufficient quality or you were captured in a compromising situation, you can send a request to the support service of the map service (Google, Yandex, Apple) indicating the coordinates and time of shooting for manual processing.
Is this data used by the police?
Officially, the companies state that the data is intended solely for navigation and cartography. However, upon lawful request by law enforcement agencies and a court order, records may be made available for crime investigations if the records capture important details of the incident.
How often are maps updated in my city?
The frequency depends on the priority of the region and the dynamics of change. Large cities can be updated once every 6-12 months, small cities - once every 1-3 years. The exact date of the last shooting can often be found in the lower corner of the screen when viewing street panoramas.