Democratizing UAV-Based Search and Rescue: Real-Time Human Detection on Affordable Edge AI Hardware

Matthew Knocker; Gabriel Goncalves Machado; Moritz Sontheimer; Shuo-Yan Chou

doi:10.3233/ATDE251079

Democratizing UAV-Based Search and Rescue: Real-Time Human Detection on Affordable Edge AI Hardware

Matthew Knocker, Gabriel Goncalves Machado, Moritz Sontheimer, Shuo-Yan Chou

National Taiwan University of Science and Technology

Research output: Contribution to conference › Paper › peer-review

Abstract

Unmanned Aerial Vehicles (UAVs) are vital in time-critical missions like search and rescue (SAR), where speed, autonomy, and adaptability are key. Advances in lightweight deep learning and low-power edge hardware now offer a path to democratize intelligent aerial systems, empowering grassroots and volunteer-led SAR efforts with affordable, open-source AI tools. This paper explores that potential by benchmarking YOLO-based object detection models on embedded platforms, including CPU-based and NPU-accelerated systems. We evaluate precision, latency, energy use, and complexity under realistic conditions, identifying model-hardware combinations that enable real-time human detection in resource-constrained environments. Our findings support a replicable, accessible framework for citizen-driven UAV deployment in disaster response

Original language	English
Pages	72-81
Number of pages	10
DOIs	https://doi.org/10.3233/ATDE251079
Publication status	Published online - 16 Oct 2025
Event	32nd International Conference on Transdisciplinary Engineering (TE2025) - EGADE Business School, Tecnologico de Monterrey, Monterrey, Mexico Duration: 7 Jul 2025 → 11 Jul 2025 https://eventos.tec.mx/s/lt-event?language=es_MX&id=a5uUG0000004r2XYAQ

Conference

Conference	32nd International Conference on Transdisciplinary Engineering (TE2025)
Country/Territory	Mexico
City	Monterrey
Period	7/07/25 → 11/07/25
Internet address	https://eventos.tec.mx/s/lt-event?language=es_MX&id=a5uUG0000004r2XYAQ

Bibliographical note

Publisher Copyright:
© 2025 The Authors.

Keywords

UAV
digital democracy
real-time small object detection
Search and Rescue
transdisciplinary engineering
community-driven technology

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.3233/ATDE251079Licence: CC BY-NC

ATDE-76-ATDE251079 (2)Final published version, 687 KBLicence: CC BY-NC

Cite this

@conference{222bead410b64544820e50834ee4bd03,

title = "Democratizing UAV-Based Search and Rescue: Real-Time Human Detection on Affordable Edge AI Hardware",

abstract = "Unmanned Aerial Vehicles (UAVs) are vital in time-critical missions like search and rescue (SAR), where speed, autonomy, and adaptability are key. Advances in lightweight deep learning and low-power edge hardware now offer a path to democratize intelligent aerial systems, empowering grassroots and volunteer-led SAR efforts with affordable, open-source AI tools. This paper explores that potential by benchmarking YOLO-based object detection models on embedded platforms, including CPU-based and NPU-accelerated systems. We evaluate precision, latency, energy use, and complexity under realistic conditions, identifying model-hardware combinations that enable real-time human detection in resource-constrained environments. Our findings support a replicable, accessible framework for citizen-driven UAV deployment in disaster response",

keywords = "UAV, digital democracy, real-time small object detection, Search and Rescue, transdisciplinary engineering, community-driven technology",

author = "Matthew Knocker and \{Goncalves Machado\}, Gabriel and Moritz Sontheimer and Shuo-Yan Chou",

note = "Publisher Copyright: {\textcopyright} 2025 The Authors.; 32nd International Conference on Transdisciplinary Engineering (TE2025) ; Conference date: 07-07-2025 Through 11-07-2025",

year = "2025",

month = oct,

day = "16",

doi = "10.3233/ATDE251079",

language = "English",

pages = "72--81",

url = "https://eventos.tec.mx/s/lt-event?language=es\_MX\&id=a5uUG0000004r2XYAQ",

}

TY - CONF

T1 - Democratizing UAV-Based Search and Rescue: Real-Time Human Detection on Affordable Edge AI Hardware

AU - Knocker, Matthew

AU - Goncalves Machado, Gabriel

AU - Sontheimer, Moritz

AU - Chou, Shuo-Yan

PY - 2025/10/16

Y1 - 2025/10/16

N2 - Unmanned Aerial Vehicles (UAVs) are vital in time-critical missions like search and rescue (SAR), where speed, autonomy, and adaptability are key. Advances in lightweight deep learning and low-power edge hardware now offer a path to democratize intelligent aerial systems, empowering grassroots and volunteer-led SAR efforts with affordable, open-source AI tools. This paper explores that potential by benchmarking YOLO-based object detection models on embedded platforms, including CPU-based and NPU-accelerated systems. We evaluate precision, latency, energy use, and complexity under realistic conditions, identifying model-hardware combinations that enable real-time human detection in resource-constrained environments. Our findings support a replicable, accessible framework for citizen-driven UAV deployment in disaster response

AB - Unmanned Aerial Vehicles (UAVs) are vital in time-critical missions like search and rescue (SAR), where speed, autonomy, and adaptability are key. Advances in lightweight deep learning and low-power edge hardware now offer a path to democratize intelligent aerial systems, empowering grassroots and volunteer-led SAR efforts with affordable, open-source AI tools. This paper explores that potential by benchmarking YOLO-based object detection models on embedded platforms, including CPU-based and NPU-accelerated systems. We evaluate precision, latency, energy use, and complexity under realistic conditions, identifying model-hardware combinations that enable real-time human detection in resource-constrained environments. Our findings support a replicable, accessible framework for citizen-driven UAV deployment in disaster response

KW - UAV

KW - digital democracy

KW - real-time small object detection

KW - Search and Rescue

KW - transdisciplinary engineering

KW - community-driven technology

UR - https://ebooks.iospress.nl/volumearticle/75473?_gl=1*1ka3am1*_up*MQ..*_ga*OTQ1OTg4MzI2LjE3NjA2MDMzNDQ.*_ga_6N3Q0141SM*czE3NjA2MDMzNDQkbzEkZzAkdDE3NjA2MDMzNDQkajYwJGwwJGgw

UR - https://pure.ulster.ac.uk/en/publications/222bead4-10b6-4544-820e-50834ee4bd03

UR - https://www.scopus.com/pages/publications/105024858734

U2 - 10.3233/ATDE251079

DO - 10.3233/ATDE251079

M3 - Paper

SP - 72

EP - 81

T2 - 32nd International Conference on Transdisciplinary Engineering (TE2025)

Y2 - 7 July 2025 through 11 July 2025

ER -

Democratizing UAV-Based Search and Rescue: Real-Time Human Detection on Affordable Edge AI Hardware

Abstract

Conference

Bibliographical note

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this