How Vehicles Determine Their Position

Introduction

Your autonomous vehicle is traveling at 80 km/h on a highway, where the system must continuously determine in real time — “Am I fully within the lane? Will I deviate in the next second?”

Looking up, the sky is overcast; looking down, the roadway is flanked by metal guardrails. Without camera perception or LiDAR returns, what enables it to maintain autonomous operation?

The answer lies in weak radio signals from satellites 20,000 kilometers above the Earth — the Global Navigation Satellite System (GNSS).

However, not all “GPS-enabled” devices are capable of meeting the requirements of L4-level autonomy.

Why can some vehicles accurately return to their lane at tunnel exits, while others “lose their way”? The difference lies in whether the system uses “standard positioning” or high-precision integrated navigation truly grounded in GNSS principles.

How GNSS Determines Position

GNSS (such as GPS and BeiDou) consists of constellations of dozens of satellites, each equipped with an atomic clock and continuously broadcasting signals containing time and orbital parameters.

The receiver determines its position using the following steps:

• Ranging: The signal propagation time from the satellite to the receiver is measured and multiplied by the speed of light to obtain the pseudorange;
• Using signals from multiple satellites: Signals from at least four satellites are used to form a system of equations, which is solved to obtain the three-dimensional position (X, Y, Z) and the receiver clock bias;
• Refinement: Carrier-phase measurements, differential corrections (e.g., RTK), and multi-frequency error mitigation techniques are used to improve positioning accuracy from meter-level to centimeter-level.

Limitations of Existing Solutions

Key Point: GNSS does not directly provide coordinates, but determines position through physical measurements and mathematical computation.

Any interference (e.g., signal blockage, multipath, and atmospheric delays) can corrupt raw measurements, resulting in positioning drift.

Most in-vehicle positioning modules have three fundamental limitations at the underlying level:

• Relying solely on pseudorange while ignoring carrier-phase measurements;
• Limited to meter-level single-point positioning, unable to support RTK, let alone achieve centimeter-level accuracy;
• A loosely coupled architecture, with GNSS and IMU operating independently, switching to inertial propagation only after GNSS signal loss, introducing discontinuities that lead to trajectory jumps;
• Unmodeled real-world error sources, lacking dynamic compensation for ionospheric delay, multipath, and satellite clock drift, leading to a sharp degradation in accuracy in urban environments.

The result? The system performs adequately in open highway conditions but begins to drift in urban environments — appearing to provide positioning, yet lacking reliability. This represents a critical risk to L4 autonomous driving.

A Tightly Coupled GNSS/INS Approach

To fully harness GNSS positioning principles, the Archimedes Innovation PBOX tightly coupled navigation receiver rebuilds the positioning engine from the ground up:

• Multi-frequency, multi-constellation full observables fusion;
• Simultaneously tracks GPS L1/L2/L5, BeiDou B1I/B2a/B3I, and Galileo E1/E5a, forming ionosphere-free combinations from triple-frequency signals to mitigate atmospheric errors;
• A tightly coupled Kalman filtering architecture;
• Fuses raw pseudorange, carrier-phase measurements, Doppler, and IMU angular velocity and acceleration in a unified state space;
• Maintains decimeter-level accuracy with as few as two satellites, constrained by the IMU;
• Position drift remains < 0.5 m within 30 seconds after carrier-phase loss;
• Real-time RTK and ambiguity resolution engine;
• Featuring a proprietary ARIA algorithm, delivering an RTK fix rate > 97% in urban canyons and a time to first fix < 8 s;
• Supports PPK and RINEX outputs;
• Records raw observables for post-processing verification, ensuring specified accuracy equals measured accuracy.

The PBOX receiver does not simply “use” GNSS — it understands it, from signal physics to estimation, delivering full control, verifiability, and reliability throughout the process.

Field Test Results (Shanghai, Urban Mixed Scenario)