AI-Driven Energy-Efficient UAV-Assisted 6G Communications: A Comprehensive Survey and Research Roadmap

Unmanned aerial vehicles (UAVs) are poised to become the key enablers of 6G—extending coverage, sensing, and edge intelligence—yet their potential remains constrained by severe energy limitations. This survey unifies the rapidly expanding field of AI-driven, energy-efficient UAV communications through a cross-layer perspective linking propulsion, radio, computation, and control. It consolidates advances in renewable-powered airframes, reconfigurable intelligent surfaces, non-orthogonal multiple access (NOMA), millimeter-wave communications, mobile edge computing, and reinforcement-learning-based coordination into a coherent analytical framework. A new taxonomy of energy costs—spanning propulsion, RF transmission, computation, sensing, and security—is introduced to benchmark mechanisms that genuinely reduce joules per bit and extend endurance. Case studies in disaster response, smart cities, precision agriculture, and rural connectivity demonstrate how trajectory–power co-optimization, learning-augmented control, and cooperative energy sharing translate theory into practice. The paper further identifies critical research gaps—including scalable swarm coordination, interference- and safety-aware learning, and reproducible energy reporting—and outlines a roadmap toward long-endurance, self-optimizing UAV–6G networks that are intelligent, efficient, and sustainable. Unlike previous reviews that examined UAV communications or 6G architectures in isolation, this work unifies the two domains through an AI-centric perspective and systematically evaluates energy-efficient design principles. This survey synthesizes more than 200 peer-reviewed works (2020–2025) across four architectural layers, offering the first bibliometric and taxonomy-based synthesis of UAV–AI–6G integration.