High-Precision Positioning (GNSS+RTK)

High-Precision Positioning (GNSS+RTK) is an augmentation technology that utilises ground base stations and differential techniques to enhance Global Navigation Satellite System (such as GPS and BeiDou) positioning accuracy from metre-level down to centimetre-level (1–10 cm).

Once confined to professional land surveying and mapping, this technology has now become a crucial technological cornerstone for autonomous driving and intelligent transport systems, enabling vehicles to pinpoint their exact positions within dedicated lanes.

💡 Core Working Principle: How Is Centimetre-Level Accuracy Achieved?

Standard standalone GNSS positioning typically yields an accuracy of around 3–10 metres, with errors primarily caused by ionospheric delay, tropospheric delay, and satellite orbit discrepancies as signals traverse the atmosphere. RTK (Real-Time Kinematic) addresses this via a "differential" approach:

1. Setting Up Base Stations: A base station is established on ground with precisely surveyed coordinates to continuously track signals from all visible satellites.

2. Computing Error Corrections: The base station compares its known physical coordinates with real-time GNSS readings to determine regional error parameters, generating "differential corrections".

3. Data Link Transmission: The base station broadcasts these corrections in real time to nearby rovers (such as autonomous vehicles) via ultra-low-latency 4G/5G networks or radio links.

4. Onboard Differential Processing: The vehicle rover receives these corrections and applies them to its own satellite raw data, effectively cancelling out most common-mode errors to deliver centimetre-level positioning accuracy.

🏙️ Real-World Challenges: Why It Cannot Be Solely Relied On

Despite RTK's impressive theoretical accuracy, it is not an all-weather silver bullet in real-world driving conditions due to several key constraints:

• Signal Blockage & Multipath Effects: In dense urban centres with towering skyscrapers (the "urban canyon" effect), road tunnels, multi-storey or basement car parks, and under heavy tree cover, GNSS signals can be easily obstructed, reflected, or attenuated. This prevents the receiver from locking onto adequate satellites or leads to heavily distorted signals, causing RTK fix failures and a drastic drop in accuracy.

• Reliance on Data Connectivity: RTK depends heavily on a seamless, low-latency data connection (typically <100 ms) to receive correction data. Any cellular drop-out immediately compromises high-precision positioning.

🚗 Its Role in Autonomous Driving: Fusion Over Isolation

Because of these limitations, GNSS+RTK rarely operates in isolation within autonomous driving architectures. Instead, it is deeply integrated with Inertial Measurement Units (IMU), Visual Odometry (VO), and LiDAR to form a robust integrated navigation suite.

• Complementary Synergies: When satellite visibility is clear, RTK serves as an absolute centimetre-level ground-truth "anchor" to eliminate long-term drift from dead-reckoning sensors. Conversely, when GNSS signals drop out temporarily (such as inside a highway tunnel), the IMU and visual/LiDAR odometry take over, delivering high-frequency, continuous relative positioning to maintain smooth tracking.

• Engineering Application: Industry applications prove that tightly coupled RTK/INS/vision fusion delivers centimetre-grade accuracy even in challenging environments while providing real-time positioning integrity monitoring. For instance, setups augmented with OBD-II wheel speed pulse data can limit positional drift to under 3.5 metres during prolonged GNSS outages in tunnels.

In summary, while GNSS+RTK provides the vital foundation for lane-level absolute positioning in autonomous vehicles, its reliance on ground infrastructure and susceptibility to environmental interference pose real challenges. Consequently, multi-sensor data fusion remains the definitive approach to guaranteeing an accurate, fail-safe, and dependable automated driving system.