The arrival of autonomous vehicles is not just about putting a car on the road without a driver, it's about achieving a extremely accurate real-time positioningReliable and continuous, even in the most challenging environments. Without it, any automated driving system falls short, no matter how good its artificial intelligence or sensor array may be.
In this context, technologies such as the Advanced GNSS, INS inertial systems, high update rates of the receivers and communication with the road infrastructure. All of this is combined with high-definition 3D maps, real-time correction networks, and connected platforms like DGT 3.0 to ensure that a vehicle knows not only where it is, but also what is happening around it and how it should react in a matter of milliseconds.
Current overview of autonomy and the need for millimeter positioning
The automotive industry is undergoing a brutal transformation, with a market of autonomous vehicles valued at hundreds of billions of dollars in the medium term and more than 500 companies worldwide developing solutions, such as Waymo boosts its technology for robotaxis.
Today, most vehicles with advanced features are located in the Level 2 automation (L2) according to SAEThat is, ADAS systems capable of maintaining lane position, adjusting speed, or following the vehicle ahead, but which require constant driver monitoring. The leap to level 3 (conditional automation) is already underway: high-end models are beginning to appear where, under certain conditions, the car manages almost everything and the driver acts as a backup, and companies like Uber will bring its robotaxis to some cities.
For these level 2, 3 and higher systems to function safely, it is not enough to simply see the road; a extremely accurate, stable, and low-latency real-time locationAt high speeds, an error of just one meter or a slow position update can be the difference between a correct maneuver and a dangerous situation, especially on highways, bridges, overpasses or complex urban environments.
Legal frameworks are also evolving. In regions such as North America, Europe, and much of Asia Regulations are being adapted to accommodate automated driving systems, robotaxis testing, and connected vehicles. All of this comes with very strict requirements regarding cybersecurity, sensor robustness, and, of course, positioning accuracy.
Sensor ecosystem: how an autonomous vehicle “sees” and “positions” itself
A modern autonomous vehicle is basically a mobile platform loaded with sensors that generate a huge amount of real-time dataA current connected car can produce around 25 GB per hour, and in fully autonomous vehicles these figures skyrocket. This constant flow of information allows the driver to "see," "hear," and "feel" the environment with a level of detail impossible for a human driver, and it is processed with specialized hardware such as NVIDIA Jetson T5000.
In terms of environmental perception, the vehicle combines different types of situational awareness sensors, a sensor fusion in software-defined vehicles:
- RGB Cameras: They provide high-resolution visual information, useful for recognizing signs, traffic lights, pedestrians, road markings, and various objects.
- LiDAR: It generates highly accurate 3D point clouds that allow the environment to be reconstructed with great geometric accuracy, ideal for high-definition maps and obstacle detection.
- Millimeter wave radar: It is key to measuring distances and relative speeds even under heavy rain, fog or low visibility, thanks to the fact that radio waves penetrate conditions where the camera or LiDAR suffers.
- Ultrasonic: They are mainly used at low speeds, in parking or very close maneuvers, to detect curbs, other vehicles or elements at very short distances.
In addition to sensing, the vehicle needs to know exactly where it is and in which direction it is moving. This is where the systems come in. GNSS/INS positioning and navigationwhich provide position, speed and orientation (heading, roll and pitch) both in open spaces and in environments where satellite signals are poor.
Without a reliable positioning backbone, the rest of the ecosystem falters. The navigation system is what allows the vehicle to determine which lane it's in, how to navigate a curve, when to change lanes, which route to follow, and how to synchronize with other vehicles and the connected infrastructure.
From traditional GPS to high-precision GNSS for autonomy
GPS revolutionized the automotive industry since it began to become more widely available in the 80s and 90s. civil and commercial useIt went from being a military system to becoming the indispensable copilot to get anywhere, optimize routes, save fuel and, in general, make life easier for drivers and fleets.
However, traditional GPS is not designed for the extreme requirements of autonomous driving. A conventional receiver typically offers position updates at 1 HzThat is, once per second. At 100 km/h, the vehicle travels almost 28 meters between two consecutive positions; at 300 km/h, more than 80 meters. This may be sufficient for basic navigation, but not for rapid maneuvers, validation of autonomous systems, or detailed analysis of vehicle dynamics.
That's why specific receivers have emerged, such as the CAN GPS sensors of 50 Hz used in automotive and autonomous vehicle testingThese devices update position, speed, and time reference 50 times per second, that is, every 20 milliseconds, capturing in much greater detail how a car moves during braking, an evasive lane change, or a curve taken at the limit.
The difference between working at 1 Hz and working at 50 Hz is similar to comparing a static photograph with a smooth videoWith a photo, you see where the vehicle was at a specific moment; with a video, you see how it approaches, brakes, turns, and exits the curve. For calibrating control algorithms, reconstructing trajectories, or synchronizing critical subsecond events, high temporal resolution is no longer an extra; it's essential.
Tightly coupled GNSS/INS navigation: the foundation of reliability
Autonomy cannot depend solely on satellite signals. In dense urban areas, tunnels, urban canyons, overpasses, or near tall buildings, the GNSS suffers from signal loss, multipath propagation, and quality degradation.If the vehicle "disappears" from the map every time it enters a tunnel, the system becomes unusable.
The solution involves tightly integrating GNSS with an inertial navigation system (INS). The INS combines accelerometers and gyroscopes to estimate position, speed, and attitude from a known starting point. On its own, INS suffers drift over time, but provides continuity when no satellites are available. GNSS, on the other hand, is stable in the long term, but can experience temporary outages.
In an architecture of tight coupling GNSS+INSThe data from both sensors are fused at the raw measurement level using a Kalman filter or other fusion algorithms. This means the system can continue calculating a navigation solution even when only a few satellites are visible, or when the signal is partially degraded, making the most of all available information.
The benefits are clear: INS fills in the gaps when GNSS is interrupted, while GNSS continuously corrects for inertial drift. The result is a much smoother, more precise and robust trajectory than with a "loose" system in which each block works on its own and only the final outputs are combined.
Specialized manufacturers like CHCNAV have developed specific GNSS+INS sensors for autonomous use, such as the series CGI-610 and CGI-830These systems integrate multi-constellation GNSS, advanced fusion algorithms, RTK support, and enhanced inertial navigation capabilities. Their specifications aim to deliver centimeter-level accuracy when corrections are available and maintain very low error rates when the GNSS signal degrades.
Accuracy and robustness requirements in real-world environments
Autonomous vehicles require a sub-meter or even centimeter positioning accuracy To know for sure which lane they are driving in, how they should position themselves at a complex intersection, or when they should overtake. It's not the same to "be on the A-1 motorway" as it is to "be in lane 3, 1,5 meters from the dividing line with lane 2."
In urban environments full of skyscrapers, the problem is compounded. The buildings block part of the satellite constellation and generate multipath reflexes which mislead the receiver, causing abrupt position jumps. To mitigate this, autonomous receivers take advantage of all available constellations (GPS, GLONASS, Galileo, BeiDou, etc.) and use real-time corrections (RTK or PPP) to reduce errors.
Another critical point is mechanical and environmental resistance. Receivers and antennas must withstand constant vibrations, sudden changes in temperature, humidity and dust without degrading its performance. A hardware failure is not just a maintenance issue: in an autonomous system it can become a direct safety risk.
Field tests under real-world conditions are essential to validate these systems. For example, in campaigns conducted in complex urban environments of Japan With high-end GNSS/INS sensors mounted on test vehicles, RMS errors of around 0,1 m have been obtained on open highways and around 0,5-1 m in very demanding urban canyons, maintaining a stable heading close to 0,1° even in long tunnels.
In tunnels, position error inevitably increases with distance traveled without GNSS, but with a good INS system, this drift is kept within acceptable limits (for example, below 1,5‰ of the tunnel length), allowing the vehicle to maintain a coherent trajectory until signal is recovered and accurately "re-engage" with the map.
Positioning at 50 Hz and validation of autonomous systems
Beyond real-time navigation for the vehicle itself, high-update-rate positioning is a key component in the development, testing and validation of autonomous vehicles and ADASTo demonstrate that a system is safe, better ground truth data than the on-board solution itself is needed.
A 50 Hz CAN GPS sensor, such as those developed by Metis Engineering, acts as that independent reference. By providing position, speed, and time data 50 times per secondIt allows for the reconstruction of trajectories with high resolution and the analysis of how the autonomous system responds in critical milliseconds, for example, in an evasive maneuver or when approaching a pedestrian crossing with moving pedestrians.
In professional motorsport, this type of sensor is used for advanced telemetry: engineers correlate the exact position on the track with variables such as brake pressure, steering angle, acceleration, or gear changesThis allows for the identification of areas where braking can be delayed, curves can be better executed, or aerodynamics can be adjusted. It's the same logic applied to autonomous vehicle testing, where every few tens of meters matters.
In the development of vehicle dynamics, the high refresh rate allows for evaluation standardized maneuvers such as rapid lane changes, slaloms, or evasive maneuvers (moose tests), analyzing how much the actual trajectory deviates from the target trajectory. This information is crucial for fine-tuning stability control algorithms, torque vectoring, or emergency assistance systems.
To enable these sensors to be easily integrated into test vehicles and industrial test benches, they rely on Automotive CAN interfaceswith power ranges compatible with 12V and 24V systems, compact and robust housings, and connectors designed to withstand vibrations, dirt and thermal changes without problems.
GNSS, INS and other high-resolution industrial positioning applications
The same technology that powers real-time positioning for autonomous vehicles is being applied to a host of fields where the trajectory, speed, and fine-tuning They are equally important. In precision agriculture, for example, automatic guidance systems need sub-decimeter accuracy to avoid overlaps and gaps in planting, fertilizing, or spraying operations.
Thanks to high update rate positioning, agricultural equipment activates and deactivates with extreme precision. fertilizer, irrigation or pesticide applicators at the boundaries of the plots, reducing waste and environmental impact. In addition, yield and soil maps align better with the reality of the terrain, making variable-rate application more reliable.
In construction and mining machinery, high-resolution GNSS systems guide excavators, graders, or transport trucks, allowing control the position of the blades and the volumes of material moved with a precision unattainable manually. This translates into less over-excavation, less additional backfill, and shorter construction times.
Mobile mapping systems mounted on land vehicles combine GNSS/INS with LiDAR, panoramic cameras, and other sensors to generate 3D maps of cities, roads, or utility networks. To ensure that each point in the LiDAR cloud is accurately georeferenced, the following are needed: navigation data at frequencies similar to those of the sensorsThis is something that specialized INS devices, such as those from OXTS, provide along with dedicated georeferencing software.
In the world of UAVs and drones, high-frequency GNSS/INS positioning improves both autonomous navigation and quality of photogrammetry and aerial surveyingThis is achieved by associating each image with a precise position and orientation, even under gusts of wind or sudden changes in trajectory. It is also critical for BVLOS (beyond line of sight) operations where the drone must follow complex routes with great accuracy.
High-definition 3D maps and their role in autonomy
Although it is sometimes said that autonomous vehicles could operate solely with their onboard sensors, in practice Detailed 3D maps remain essentialHaving a pre-existing map allows the perception system to offload some of its work and makes it easier for the algorithms to focus on detecting what changes: pedestrians, other cars, construction, unexpected obstacles, etc.
These three-dimensional maps include not only the geometry of the road, but also buildings, trees, signs, traffic lights, street furniture and even details such as lane width or the superelevation of curves. Specialized companies, such as TomTom and various mobile mapping firms, invest heavily in generating and maintaining this high-definition mapping.
To build them, mapping vehicles equipped with LiDAR, cameras and high-precision GNSS/INS systemsThe INS provides the attitude (roll, pitch, yaw), and the GNSS, supported by differential corrections, determines the absolute position. The result is dense and precise point clouds that are then processed into classified point clouds, surface models, or maps for autonomous navigation.
The final quality of these maps depends directly on the quality of the integrated positioning system. It's not very useful to have an excellent LiDAR system if it can't be used effectively. georeference each point with sufficient accuracyThat's why INS devices for mobile mapping include advanced filters, tight integration with GNSS, and post-processing tools that correct even small deviations using reference station data.
Ultimately, the autonomous vehicle positions itself in real time by comparing what it sees with what the map indicates should be there. If there is a significant discrepancy (for example, a street closed for construction that wasn't shown on the map), the system adjusts its route planning, alerts the user, and, in some cases, sends update information to the platform so that this change can be incorporated into future maps.
Connected mobility, DGT 3.0 and geolocated V16 beacons
The story doesn't end with the vehicle itself. In Spain, the DGT (Spanish Directorate General of Traffic) is promoting the platform DGT 3.0, which seeks to transform roads into smart environments where cars, infrastructure and safety devices communicate with each other in real time.
One of the key elements in this ecosystem is the V16 emergency beacon with geolocationwhich will definitively replace emergency triangles from 2026. These beacons, offered by manufacturers such as SOOS, are placed on the roof of the broken-down vehicle and, in addition to emitting a flashing light visible from a great distance, send their position to the DGT cloud using NB-IoT or LTE-M technologies.
This automatic transmission of coordinates allows DGT 3.0 to share the location of the immobilized vehicle with others connected users, support services and, in the future, autonomous vehiclesThus, a car with V2X capability could receive an alert that there is a vehicle stopped a few hundred meters ahead and adapt its speed or trajectory in advance.
We are still in a transition phase: today there is no fully standardized protocol for any autonomous vehicle to communicate directly with any approved V16 beacon, but V2I (vehicle-to-infrastructure), V2V (vehicle-to-vehicle), and V2N (vehicle-to-network) technologies are paving the way. The goal is an ecosystem where road safety devices, infrastructure, and vehicles interact. engage in continuous and secure dialogue.
Significant challenges remain: lack of protocol standardization, interoperability between manufacturers, data transmission latency, and legal issues. Privacy and protection of geolocation dataEven so, the role of brands like SOOS, which already offer beacons prepared for this connected future with connectivity included for many years, accelerates the adoption of this type of solution.
All this network of advanced GNSS, high-quality INS, 50 Hz positioning, detailed 3D maps, and connected devices like V16 beacons with geolocation is converging towards a single goal: to ensure that autonomous vehicles have access to a constant, real-time location. a clear, accurate and up-to-date vision of where they are, what surrounds them and how they should actAs standards are refined, legal frameworks are strengthened, and technology continues to mature, real-time positioning is consolidating itself as one of the most critical pillars on which the autonomous and connected mobility of the near future is built.
