Quantum Reservoir Computing for Noisy, High-Dimensional Magnetic Navigation

Magnetic navigation is attractive in GPS-denied or GPS-degraded environments, yet real-world magnetic measurements are commonly noisy and high-dimensional, limiting the effectiveness of conventional filtering and model-based approaches. We address this challenge with quantum reservoir computing (QRC), which projects inputs into an exponentially large Hilbert space whose native dynamics provide rich memory and nonlinear mixing while retaining a simple linear readout. In our framework, high-dimensional magnetic features are injected into the quantum reservoir, and a single linear layer is trained to perform denoising, yielding robust noise suppression directly on raw sensor streams. Beyond an application demonstration, our results highlight QRC as a practical tool for learning-based filtering in high-dimension, noisy sensing pipelines, offering a compact training footprint with generalization in GPS-challenged settings.