Reinforcement Learning for Adaptive Flight Envelope Protection in Fly-by-Wire Systems
Proposes a reinforcement learning controller that adapts flight envelope protection limits in real time under degraded aerodynamic conditions.
Search across all published articles, or filter by topic area.
Proposes a reinforcement learning controller that adapts flight envelope protection limits in real time under degraded aerodynamic conditions.
Surveys 52 studies applying machine learning to short- and medium-term conflict prediction in air traffic management.
Presents a monocular vision pipeline for runway detection and glideslope estimation, reducing lateral touchdown error by 38% relative to an inertial-only baseline.
Combines vibration, thermal, and exhaust gas temperature sensor streams in a multivariate LSTM model to forecast turbofan component degradation ahead of scheduled maintenance windows.
Describes a digital twin pipeline synchronizing strain gauge telemetry with a finite-element model to estimate real-time fatigue accumulation across primary airframe structures.
Evaluates a speech-to-text and intent verification pipeline for automatically flagging readback discrepancies in pilot-controller communication, tested against 4,000 recorded transmissions.
Proposes a decentralized coordination protocol for multi-UAV search patterns under intermittent communication, validated in a simulated mountainous terrain environment.
Compares attention-based explanation methods for a diversion-recommendation model, measuring pilot trust and decision time in a fixed-base simulator study with 24 commercial pilots.
Applies federated learning across four partner airline fleets to train a shared component failure model without centralizing proprietary maintenance logs.