Toward Field-Ready Quantum Sensing: Autoencoding for Robust Operation

Quantum sensors deliver exceptional laboratory sensitivity but suffer performance losses in operational settings due to vibrations, thermal drift, and electromagnetic interference (EMI). We introduce a novel deployable quantum sensing framework that trains an autoencoder with physics-informed objectives on controlled disturbance injections and deploys only the decoder for causal disturbance estimation, uncertainty quantification, and control adaptation. The decoder maps auxiliary channels to a disturbance estimate that is subtracted from the quantum readout, yielding real-time suppression with strict latency and safety constraints. New theorems (deferred to the Appendix) establish the theoretical foundations of the proposed framework, proving its decoderbased approximation accuracy, fundamental noise rejection limit, calibrated uncertainty, latency–performance trade-off, and both computational and quantum scalability.