Development of an Edge-AI Autonomous UAV Control System for Navigation and Perception in GPS-Denied Environments

Abstract

In aerospace research on the cross-cutting growth of unmanned aerial vehicles (UAVs) in inspection, monitoring, and emergency response use cases, reliable operation in environments that are constrained by navigation and communication infrastructure is an issue of vital importance. Conventional UAV systems have a strong reliance on Global Positioning System (GPS) signals and constant telemetry with ground control stations that are often the cause of performance degradation when operational inside indoor environments, dense urban canyons and industrial complexes. In view of these troubles, in the present work, an edge artificial intelligence (AI) based autonomous UAV system is proposed to accomplish sensing, localization, and mission control processes fully onboard; hence nullifying the requirement of external aids for navigation. The proposed methodology works on GPS-free navigation by merging the information from the visual and inertial sensors for the accurate estimation of position in real time. Environmental awareness is achieved through object recognition procedures implemented on board, while a dedicated autonomous mission control module is responsible for the execution of high-level tasks within strict and defined safety bounds. All the computational routines reside in an embedded computing platform, ensuring low-latency responses and longevity under degraded communication scenarios. Empirical validation of the proposed system was achieved by experiments of extensive indoor and outdoor flight in a GPS-denied environment. The results show an 84\% completion rate for a fully autonomous mission, in addition to stable flying behaviour and timely control responses, confirming the feasibility of reliable UAV autonomy without reliance on external navigation infrastructure.