Precise and robust localization for micro Unmanned Aerial Vehicles (UAVs) in GNSS-denied environments is hindered by the lack of diverse datasets and the limited real-world performance of existing visual matching methods. To address these gaps, we introduce two contributions: (1) the SatLoc dataset, a new benchmark featuring synchronized, multi-source data from varied real-world scenarios tailored for UAV-to-satellite image matching, and (2) SatLoc-Fusion, a hierarchical localization framework. Our proposed pipeline integrates three complementary layers: absolute geo-localization via satellite imagery using DinoV2, high-frequency relative motion tracking from visual odometry with XFeat, and velocity estimation using optical flow. An adaptive fusion strategy dynamically weights the output of each layer based on real-time confidence metrics, ensuring an accurate and self-consistent state estimate. Deployed on a 6 TFLOPS edge computer, our system achieves real-time operation at over 2 Hz, with an absolute localization error below 15 m and effective trajectory coverage exceeding $90\%$, demonstrating state-of-the-art performance. The SatLoc dataset and fusion pipeline provide a robust and comprehensive baseline for advancing UAV navigation in challenging environments.

The Vortex Ring State (VRS) poses a catastrophic aerodynamic threat to coaxial dual-rotor unmanned aerial vehicles (UAVs). Traditional reactive detection mechanisms provide insufficient altitude for recovery, while existing data-driven diagnostics are severely bottlenecked by data leakage, extreme class imbalance, and a lack of physical interpretability. To bridge these gaps, this paper proposes a physics-informed multimodal deep learning framework that transitions from post-occurrence detection to proactive early warning. We establish a 1.5 s precursor window—creating a three-class ordinal state space—to provide the flight control system with critical intervention time for differential rotor recovery. We developed a novel ViT-LSTM architecture (MTSF-Net) to fuse continuous seven-channel onboard-recorded data (comprising three-axis acceleration, three-axis angular velocity, and barometric vertical velocity), which are subsequently transformed into Continuous Wavelet Transform (CWT) spectrograms. To ensure real-time unidirectional inference while preserving absolute physical vibration scales across heterogeneous sensors, a Calibrated Benchmark Normalization (CBN) strategy is introduced. Furthermore, a Hybrid Ordinal Loss is proposed to mitigate the extreme sample imbalance (<0.5%) of the precursor state by penalizing asymmetric aerodynamic degradation. Evaluated under a strict sortie-based isolation protocol, the proposed system achieves an exceptional test accuracy of 98.26% and an unprecedented precursor recall of 100%. Notably, it completely eliminates fatal missed detections (VRS predicted as Normal) and false-positive VRS predictions triggered by precursor states. Finally, Gradient-weighted Class Activation Mapping (Grad-CAM) is utilized to verify that the multimodal sensor processing pipeline successfully anchors onto authentic physical vibration frequencies rather than artifactual noise, laying a rigorous, interpretable foundation for intelligent aviation safety systems.

The Vortex Ring State (VRS) is a critical aerodynamic hazard for rotorcraft, characterized by highly unsteady fluid–structure interactions and severe low-frequency vibrations. While data-driven deep learning models have shown promise in aviation state monitoring, their inherent “black-box” nature fundamentally contradicts the stringent interpretability requirements of airworthiness certification. To address this, we propose an “AI for Science” paradigm, investigating whether advanced Vision Transformers (ViT) can autonomously discover underlying aerodynamic mechanisms without human physical priors. First, to ensure absolute data fidelity, flight test datasets of a coaxial unmanned aerial vehicle were rigorously labeled using cross-validation from high-fidelity Computational Fluid Dynamics (CFD) simulations and wind tunnel tests. One-dimensional vibration signals were then transformed into two-dimensional Continuous Wavelet Transform (CWT) spectrograms. By employing Target-Layer Gradient Adaptation (Grad-CAM) techniques, we conducted a systematic comparison between traditional Convolutional Neural Networks (ResNet50) and ViT. The results demonstrate that while CNNs suffer from diffuse attention caused by high-frequency noise, the frozen-backbone ViT model achieves a physically interpretable accuracy of $93.24\%$, while autonomously locking its global attention onto a perfectly horizontal feature band centered at $41.7$ Hz. Crucially, this autonomously discovered feature precisely aligns with the theoretically derived once-per-revolution (1P) fundamental frequency of the rotor’s flap-lag coupling response under VRS aerodynamic turbulence. This research provides direct visual evidence bridging black-box AI decisions with classical fluid mechanics, proposing a “Mechanism-Guided Verification” framework that offers a trustworthy pathway for the future certification of AI in safety-critical aerospace systems.