The Role of GNSS in Autonomous Driving Systems
· ⏱ 11 min read · 👁 viewsShare
From robotaxis and autonomous delivery vehicles to port automated guided vehicles (AGVs) and self-driving mining trucks, autonomous driving technology is expanding into an increasing range of real-world applications. For an autonomous vehicle, identifying roads, vehicles, and obstacles is only part of the challenge. It must also continuously determine its position, heading, and motion, while assessing whether its current localization solution is reliable enough for safe operation.
Global Navigation Satellite Systems (GNSS) provide absolute position, velocity, and precise timing information within a globally consistent reference framework. This makes GNSS an important component of autonomous vehicle localization. However, real-world driving environments rarely offer ideal satellite visibility. Buildings, trees, tunnels, multipath, and radio-frequency (RF) interference can all degrade GNSS positioning performance.
As a result, modern autonomous driving systems are moving beyond centimeter-level accuracy alone. By integrating GNSS with Inertial Measurement Units (IMUs), cameras, LiDAR, and other sensors, autonomous vehicles can build more robust localization systems designed to improve accuracy, availability, continuity, and reliability.
This article explores the role of GNSS in autonomous driving, the importance of GNSS/INS integration, the challenges of reliable vehicle localization, and how GNSS technology supports different autonomous driving applications.
1. Why GNSS Is Essential for Autonomous Vehicle Localization
Conventional vehicle navigation generally needs to determine which road a vehicle is traveling on. Autonomous driving requires much finer localization. The system may need to determine which lane the vehicle occupies, how far it is from a lane or road boundary, and how closely its actual trajectory follows the planned path.
Standard GNSS standalone positioning typically provides meter-level accuracy. When combined with high-precision techniques such as RTK (Real-Time Kinematic), carrier-phase observations and correction data can be used to achieve centimeter-level positioning. Under favorable satellite visibility and correction conditions, a high-precision GNSS receiver can achieve horizontal positioning accuracy of approximately 8 mm + 1 ppm and vertical accuracy of approximately 15 mm + 1 ppm, providing a more precise absolute reference for lane-level localization, path planning, and vehicle control.
However, GNSS does much more than simply output coordinates. GNSS receivers can also provide precise timing information and establish a common time reference through signals such as PPS (Pulse Per Second). Autonomous vehicles typically operate multiple sensors simultaneously, including cameras, LiDAR, IMUs, and other onboard devices. For effective sensor fusion, measurements from these sensors must be correctly aligned in time. Consider a vehicle traveling at 72 km/h, or approximately 20 m/s. If there is a 50 ms timing offset between two sensors, the vehicle will have moved about one meter during that interval. Even if each sensor is highly accurate on its own, incorrect time synchronization can introduce significant errors into the fused solution. Dual-antenna GNSS can also provide vehicle heading. By measuring the carrier-phase relationship between two GNSS antennas installed with a known baseline, the receiver can estimate heading without relying entirely on the vehicle's direction of motion. This capability is particularly valuable for low-speed vehicles, mining trucks, AGVs, and other large autonomous platforms. GNSS therefore serves several important functions in autonomous driving, providing a reference for absolute position, velocity, time, and heading.

2. Why GNSS/INS Integration Is Essential for Autonomous Driving
One of the greatest advantages of GNSS is its ability to provide a long-term absolute position reference without the accumulated drift associated with inertial navigation. At the same time, GNSS has an inherent limitation: it depends on radio signals transmitted from satellites.
Under open-sky conditions, a receiver can normally track a large number of satellites and maintain good positioning performance. In urban canyons, however, buildings may block direct satellite signals. Signals reflected by buildings, glass façades, and other surfaces can also create multipath and NLOS (Non-Line-of-Sight) measurements, reducing positioning accuracy. In tunnels and underground parking facilities, GNSS signals may become completely unavailable. Complex RF environments can also introduce narrowband, broadband, or continuous-wave interference that affects satellite signal acquisition and tracking. For these reasons, modern autonomous vehicles generally do not treat GNSS as their only source of localization. Instead, they may use a multi-sensor architecture such as: GNSS + IMU + Camera/LiDAR + Map. Within this architecture, GNSS provides a long-term absolute position reference, while the IMU continuously measures vehicle motion using accelerometers and gyroscopes. Because an IMU does not depend on external radio signals, it can continue operating even when a vehicle enters a tunnel or another GNSS-denied environment.
The two technologies are naturally complementary. GNSS does not suffer from the accumulated integration drift of inertial navigation, but its performance is strongly influenced by the surrounding signal environment. An IMU, on the other hand, can provide continuous motion information at 100 Hz or even higher rates, but small sensor biases and noise accumulate over time and gradually increase the navigation error. With GNSS/INS integration, reliable GNSS observations can be used to estimate and constrain inertial navigation errors. When GNSS is temporarily blocked or interrupted, the INS can continue propagating the vehicle's position, velocity, and attitude. Once reliable GNSS measurements become available again, they can be used to constrain the accumulated inertial error. Adding cameras, LiDAR, and high-definition maps provides further constraints based on road features, lane markings, buildings, and point-cloud information. The purpose of multi-sensor fusion is therefore not to determine which sensor is “best.” It is to allow different sensors to complement one another under the conditions in which each performs most effectively.
3. Beyond Accuracy: Positioning Integrity and Reliability
For conventional surveying applications, positioning accuracy is often one of the primary performance indicators. In autonomous driving, however, achieving centimeter-level accuracy does not automatically mean that the localization solution is sufficiently reliable. A more important question is: Does the system know whether its current positioning solution can still be trusted? This is closely related to the concept of positioning integrity. GNSS receivers can provide information such as the number of tracked satellites, C/N₀, DOP values, positioning status, and RTK solution status. Some systems can also use techniques such as RAIM (Receiver Autonomous Integrity Monitoring), FDE (Fault Detection and Exclusion), and other consistency checks to identify potentially abnormal satellite measurements.
For example, a GNSS receiver may continue to output a high-precision position after a vehicle enters a dense urban environment. However, changes in satellite geometry, increased multipath, NLOS reception, or RF interference may reduce the actual reliability of that solution. An autonomous localization system therefore needs to continuously evaluate GNSS quality rather than simply passing every position output directly to the vehicle control system. When GNSS quality deteriorates, the fusion algorithm can reduce the weight assigned to GNSS and rely more heavily on INS, vision, LiDAR, or map matching. If significant inconsistencies appear between different localization sources, the system can perform fault detection or transition to an appropriate degraded operating mode.
This is one of the key ideas behind localization redundancy. Autonomous driving also places demanding requirements on GNSS hardware itself, including high data output rates, interference mitigation, rapid signal reacquisition, and the ability to operate reliably under automotive temperature, vibration, and long-term operating conditions. For example, the Qtalis M20 X GNSS/INS receiver is designed for automotive applications and supports interfaces including Ethernet, CAN/CAN FD, and RS232. By integrating GNSS with a high-performance IMU, it can serve as one of the positioning sources within a multi-sensor autonomous driving architecture.
The focus of autonomous localization is therefore expanding beyond High Accuracy toward four broader objectives: Accuracy + Availability + Continuity + Integrity. In other words, a positioning system should not only provide an accurate position. It should also remain available and continuous whenever possible, while providing enough information for the system to recognize when the solution may no longer be reliable.
4. GNSS Applications in Autonomous Vehicles
Robotaxis, autonomous delivery vehicles, port AGVs, and self-driving mining trucks all fall under the broad category of autonomous vehicles, but they operate in very different GNSS environments. Their localization systems therefore have different priorities.
Robotaxis and Robobuses primarily operate on complex urban roads. Tall buildings, overpasses, tree-lined streets, and constantly changing satellite visibility make it difficult for GNSS alone to maintain consistently high positioning performance. These systems therefore commonly combine GNSS/INS with cameras, LiDAR, and high-definition maps. GNSS/INS receivers such as the M20 X can communicate with vehicle domain controllers through Ethernet, CAN/CAN FD, and other interfaces, providing absolute position and inertial information to the overall localization system.
Autonomous delivery vehicles often operate in campuses, industrial parks, residential communities, and similar environments. Although their operating speeds are relatively low, buildings and trees can still cause significant satellite signal obstruction. These battery-powered platforms may also place greater emphasis on power consumption and device size. The Qtalis K901, which supports GNSS/INS integrated navigation with low power consumption, can be used in autonomous platforms where size and power efficiency are important considerations.
Port AGVs face a different set of challenges. Ports may contain large machinery, communication equipment, and complex electromagnetic environments. In these applications, interference mitigation can be just as important as centimeter-level positioning. The Qtalis K902 supports adaptive interference mitigation techniques that can suppress certain narrowband and continuous-wave interference, helping improve GNSS availability in challenging RF environments.
Autonomous mining trucks must operate in demanding conditions involving rough terrain, vehicle vibration, large equipment, changing satellite visibility, and dynamic motion. Stable heading information is also important for these large vehicles. The Qtalis K922 supports dual-antenna heading and GNSS/INS integrated navigation. Under suitable installation and satellite observation conditions, it can provide centimeter-level positioning together with heading accuracy of approximately 0.1°/R (R refer to Baseline, unit: m).
Selecting a GNSS solution for autonomous driving therefore involves much more than comparing a single positioning accuracy specification. Satellite visibility, vehicle dynamics, interference conditions, INS performance, interfaces, power consumption, and system redundancy should all be considered as part of the overall localization architecture.
5. The Future of Autonomous Localization: GNSS as Part of a Multi-Sensor System
Future autonomous localization systems are unlikely to focus on centimeter-level accuracy alone. Development is increasingly moving toward wider coverage, faster convergence, deeper sensor fusion, integrity monitoring, and functional safety.
PPP-RTK is one technology attracting growing attention. It combines characteristics of wide-area PPP correction services with fast high-precision positioning approaches. By using information such as satellite orbit and clock corrections, signal biases, and atmospheric corrections, PPP-RTK can support high-precision positioning over larger service areas.
Satellite-based augmentation and high-precision correction services are also expanding the ways in which precise GNSS positioning can be delivered. Services such as Galileo HAS, BeiDou high-precision services, and QZSS CLAS can provide correction information through satellite broadcasts, which can be particularly valuable in areas where terrestrial communication coverage is limited.
Another major trend is deeper integration among GNSS, IMUs, cameras, and LiDAR.
Future localization systems may go beyond simply combining the final position outputs of several independent sensors. Raw GNSS measurements, inertial measurements, visual features, and LiDAR point clouds can instead be integrated through tightly coupled or deeply integrated estimation architectures.
When GNSS is affected by multipath or signal blockage, visual and LiDAR information can provide additional constraints. When cameras are affected by darkness, strong sunlight, or adverse weather, GNSS and INS can continue to provide vehicle motion information. When one positioning source behaves abnormally, other sensors can also provide independent information for consistency checking and fault detection.
6. Conclusion:
The central challenge for future autonomous localization is therefore no longer simply:“Can we achieve centimeter-level positioning?”Instead, the question is: “Can we continuously provide accurate, available, and reliable positioning in changing real-world environments while also knowing when the solution should no longer be trusted?”
GNSS will not perform every localization task in an autonomous vehicle on its own. However, as a globally consistent source of absolute position and precise time, it remains an important foundation of the multi-sensor localization system. From RTK positioning and GNSS/INS integration to dual-antenna heading, interference mitigation, and integrity monitoring, GNSS is evolving from a conventional positioning technology into an important component of reliable autonomous driving localization.
📘 Recommended Reading
When GNSS Is Lost: How Does the IMU Keep Navigation Alive?
Learn how GNSS/INS systems use IMU measurements to maintain navigation during GNSS outages, why inertial drift occurs, and how navigation recovers when GNSS signals return.
Continue the Conversation
Have questions about this topic or want to share your experience? Join the Qtalis Community to discuss GNSS technologies, exchange ideas, and connect with other users and developers.
Join the Discussion