Porting Computer Vision Models to the Edge for Smart City Applications: Enabling Autonomous Vision-Based Power Line Inspection at the Smart Grid Edge for Unmanned Aerial Vehicles (UAVs)

Date
2022-01-04
Authors
Gudmundsson , Ingi
Falco, Gregory
Contributor
Advisor
Department
Instructor
Depositor
Speaker
Researcher
Consultant
Interviewer
Annotator
Journal Title
Journal ISSN
Volume Title
Publisher
Volume
Number/Issue
Starting Page
Ending Page
Alternative Title
Abstract
Smart grid infrastructure must be monitored and inspected - especially when subject to harsh operating conditions in extreme, remote environments such as the highlands of Iceland. Current methods for monitoring such critical infrastructure includes manual inspection, static video analysis (where connectivity is available) and unmanned aerial vehicle (UAV) inspection. UAVs offer certain inspection efficiencies; however, challenges persist given the time and UAV operator skill required. Collaborating with Landsnet, the Icelandic smart grid operator, we apply convolutional neural networks for image processing to detect smart grid transmission infrastructure and modify the resulting computer vision (CV) model to function on the edge of a UAV. In doing so, we overcome significant edge processing barriers. Our real-time CV model delivers decision insight on the UAV edge and enables autonomous flight path planning for use in smart grid inspection. Our approach is transferable to other smart city applications that could benefit from edge-based monitoring and inspection.
Description
Keywords
Smart (City) and Data Streaming Application Development: Challenges and Experiences, autonomous uavs, computer vision, edge devices, infrastructure monitoring, smart grid
Citation
Extent
10 pages
Format
Geographic Location
Time Period
Related To
Proceedings of the 55th Hawaii International Conference on System Sciences
Table of Contents
Rights
Attribution-NonCommercial-NoDerivatives 4.0 International
Rights Holder
Local Contexts
Email libraryada-l@lists.hawaii.edu if you need this content in ADA-compliant format.