Quantum Position Navigation Timing for Autonomous Vehicles: A Federated Learning Framework

Global Navigation Satellite System (GNSS) outages–caused by blockage, multipath, or interference–remain a fundamental obstacle to autonomous-vehicle operation. Classical inertial navigation provides short-term continuity but suffers unbounded drift in unaided, long-duration GNSS denial. This paper proposesQPNT-FL, a privacy-preserving framework that couples quantum Position, Navigation, and Time (QPNT) sensing with federated learning (FL) to achieve drift-bounded navigation without centralizing sensitive trajectories. Each vehicle fuses heterogeneous quantum modalities (matter-wave inertial phases, atomic frequency references, and gravimetric/magnetic field measurements) using a modular encoder–fusion network with cross-attention, while FL continuously refines calibration and fusion parameters across the fleet under differential privacy. In a fully GNSS-denied hours of experiments with vehicles operating across urban canyons, tunnels, mountainous electromagnetically contested regions, QPNT-FL maintains a bounded steady-state positioning and achieves error reduction.