A Conceptual Architecture for Resilient UAV Navigation in GNSS-Degraded Environments

Abstract

Global Navigation Satellite System (GNSS) signals are often unreliable or unavailable in urban canyons, dense forests, or indoor environments, posing significant challenges for UAV navigation. This paper proposes a conceptual architecture for resilient UAV navigation that integrates multi-sensor fusion, adaptive filtering, and environment-aware localization strategies to maintain accurate state estimation under GNSS-degraded conditions. The architecture is modular, combining inertial measurement units (IMUs), vision-based odometry, LiDAR, and barometric altimeters within a hierarchical fusion framework. We introduce the concept of “sensor trust modulation,” where the relative confidence in each sensor modality dynamically adjusts based on environmental context and sensor health diagnostics. The architecture also incorporates a fault detection and isolation (FDI) layer to identify and mitigate sensor anomalies in real-time. Theoretical foundations are laid for observability analysis in multi-sensor fusion systems, emphasizing the importance of persistent excitation and redundancy. By formalizing the interplay between sensor modalities and environmental constraints, this framework aims to guide the design of robust navigation systems capable of seamless operation across diverse mission profiles. The paper concludes with a discussion on the implications for autonomous mission planning and the potential for integration with emerging technologies such as event-based cameras and deep learning-based localization.