Edge Intelligence in Autonomous UAV Power Line Inspection: Application Modes and Technical Challenges
Abstract
Traditional Unmanned Aerial Vehicle (UAV) power line inspection largely relies on a "store-and-forward model, which suffers
from high latency and dependency on stable communication networks, particularly in remote environments. To address these limitations, this
paper explores the integration of Edge Intelligence (EI) into UAV systems to enable real-time, analysis-as-you-fly capabilities. We first analyze typical application modes, including real-time onboard defect recognition and dynamic flight path replanning based on situational awareness. Subsequently, the paper identifies critical technical challenges regarding onboard computing constraints, energy consumption, and the
deployment of complex deep learning models. To overcome these hurdles, we review state-of-the-art solutions ranging from specialized AI
hardware acceleration to algorithm lightweighting techniques and federated learning. Finally, the paper envisions a future "end-edge-cloud
collaborative paradigm to enhance the autonomy and efficiency of power line inspections.
from high latency and dependency on stable communication networks, particularly in remote environments. To address these limitations, this
paper explores the integration of Edge Intelligence (EI) into UAV systems to enable real-time, analysis-as-you-fly capabilities. We first analyze typical application modes, including real-time onboard defect recognition and dynamic flight path replanning based on situational awareness. Subsequently, the paper identifies critical technical challenges regarding onboard computing constraints, energy consumption, and the
deployment of complex deep learning models. To overcome these hurdles, we review state-of-the-art solutions ranging from specialized AI
hardware acceleration to algorithm lightweighting techniques and federated learning. Finally, the paper envisions a future "end-edge-cloud
collaborative paradigm to enhance the autonomy and efficiency of power line inspections.
Keywords
Edge Intelligence; UAV Power Line Inspection; Real-time Defect Detection; Lightweight Neural Networks; Model Compression;
End-Edge-Cloud Collaboration
Full Text:
PDFReferences
[1] Liu Z, et al. Unmanned aerial vehicles and edge AI for power line inspection [J]. Remote Sensing, 2023, 15(2): 539.
[2] Huang X, Xie W, Zhang Q, et al. TLI-YOLOv5: A lightweight object detection framework for transmission line inspection [J].
Electronics, 2023, 12(15): 3340.
[3] Xu Q, Zhang Y, Peng H, et al. Cloud-edge-end collaboration for power line inspection: a proximal policy optimisation-based image
offloading approach [J]. Engineering Research Express, 2025, 7(4): 045313.
DOI: http://dx.doi.org/10.70711/aitr.v3i6.8594
Refbacks
- There are currently no refbacks.
