The DOI matches its Crossref record; 155 of the 163 references it cites were found in a registry or an index; 14 of the 83 sources its claims cite were checked for support.
A Survey on the Convergence of Edge Computing and AI for UAVs: Opportunities and Challenges
Open accessAbstract
The latest 5G mobile networks have enabled many exciting Internet of Things (IoT) applications that employ unmanned aerial vehicles (UAVs/drones). The success of most UAV-based IoT applications is heavily dependent on artificial intelligence (AI) technologies, for instance, computer vision and path planning. These AI methods must process data and provide decisions while ensuring low latency and low energy consumption. However, the existing cloud-based AI paradigm finds it difficult to meet these strict UAV requirements. Edge AI, which runs AI on-device or on edge servers close to users, can be suitable for improving UAV-based IoT services. This article provides a comprehensive analysis of the impact of edge AI on key UAV technical aspects (i.e., autonomous navigation, formation control, power management, security and privacy, computer vision, and communication) and applications (i.e., delivery systems, civil infrastructure inspection, precision agriculture, search and rescue (SAR) operations, acting as aerial wireless base stations (BSs), and drone light shows). As guidance for researchers and practitioners, this article also explores UAV-based edge AI implementation challenges, lessons learned, and future research directions.
Overview
CoChat Summary
TL;DR
This survey examines the convergence of edge computing and AI for UAVs, arguing that edge AI can better meet UAV requirements for low latency, reliability, energy efficiency, security, and privacy than traditional cloud AI. It reviews six UAV technical challenges, six UAV applications, implementation challenges, lessons learned, and future research directions.
Key findings
- Edge AI is defined as “the capability to enable edges to execute AI algorithms”; it inherits edge computing benefits (lower latency, higher reliability, improved security/privacy, reduced cost, reduced energy consumption) and adds benefits such as federated learning, further energy savings, and aggregation frequency control.
- The survey identifies six key UAV technical challenges—autonomous navigation, formation control, power management, security and privacy, computer vision, and communication—and six key applications—delivery systems, civil infrastructure inspection, precision agriculture, search and rescue, acting as aerial wireless BSs, and drone light shows.
- Reported figures include: the commercial UAV market was estimated at U.S. $20.8 billion in 2021 and expected to reach U.S. $501.4 billion by end of 2026; the drone market is expected to grow at a 57.5% compound annual growth rate from 2021 to 2028; construction and infrastructure inspection applications take up about 45% of the total UAV market; a Guinness world record drone light show used 3051 drones in Guangdong, China, September 2020; Qualcomm claims edge AI-optimized chips can yield energy savings as much as 25× relative to conventional chips; battery swapping time is roughly 60 s versus average battery charging time of 45 to 60 min.
- Implementation challenges for UAV edge AI include developing distributed training algorithms, security and privacy, resource allocation, and real-time requirements; other challenges include making statically trained models more adaptable, generalizing reliably over unseen data, accurately evaluating runtime performance at the edge, and improving synchronization among UAV, edge, and cloud.
Methods
- Design: A survey article providing a comprehensive qualitative analysis of edge AI for UAVs, with sections on background, technical challenges, applications, implementation challenges, lessons learned, and future directions.
- Data or sample: No primary empirical sample; the work synthesizes and compares prior literature, including related surveys and cited studies, and uses summary tables such as Table I comparing related surveys and Tables III–VI summarizing challenges, applications, requirements, and implementation challenges.
- Main analysis: Qualitative categorization and synthesis of how edge AI affects UAV technical challenges and applications, plus discussion of possible solutions and remaining research questions.
Limitations
- Stated in the paper: edge AI disadvantages include significantly fewer network devices in some places and fewer skilled people to implement, fix, or manage network devices in such places; task offloading between different edge servers can limit UAV mission performance due to high UAV speeds; distribution of data processing increases potential attack surface; blockchain-based UAV swarms face 51% attacks if over half the UAVs are hacked; federated learning can be vulnerable to membership inference attacks when the adversary is a participant; costs of launching and maintaining edge devices spread over large areas can be significant; embedded computer vision is limited by onboard computational power; and edge AI helps drone light shows in a more limited capacity because UAVs in such shows generally communicate directly with a GCS.
- Evident from the text: as a survey, it does not present new experiments or quantitative validation of edge AI performance on UAVs; its findings are synthesized from cited literature, and some discussions are high-level or rely on cited works rather than original measurements.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
14 Valid · 8 Not valid · 68 Not checked · 73 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“In the case of training a deep learning model in an edge computing environment, where distributed models are trained locally first and then updates are aggregated centrally (e.g., federated learning), the control of updates aggregation frequency has a significant influence on the communication overhead.”
§ A Survey on the Convergence of Edge Computing and AI for UAVs: Opportunities and Challenges > 3) Further Reduce Communication Overhead in Certain Scenarios by Enabling Aggregation Frequency Control:
What the sources say
- Supportedcites [8] source
The “Aggregation Frequency Control” section states that edge-training models may be trained locally and their updates aggregated centrally, as in federated learning. It explicitly says aggregation frequency significantly influences communication overhead. No page number is supplied in the copy. (LLM, conf=0.95)
- Not matchedcites [40] Hsieh, Gaia: Geo-distributed machine learning approaching {LAN$} speeds
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Additionally, when edge AI is employed, this sending of data to servers and the energy consumption associated with it can be further decreased through distributed federated learning within a UAV swarm.”
§ A Survey on the Convergence of Edge Computing and AI for UAVs: Opportunities and Challenges > C. Power Management 1) Introduction:
What the sources say
- Supportedcites [88] source
The Introduction (p. 1) states that centralized ML is challenging for UAV swarms because it requires transmitting a large volume of data over aerial links to ground base stations, and that with distributed FL 'UAVs do not need to send any raw data to BSs when training learning models' — i.e., the data offloading to centralized servers is eliminated by FL within the swarm. The same paper explicitly treats the energy side of this exchange: the Federated learning model section (p. 3) computes the training/transmission energy E_L and E_i from the model packet sizes S(w_i), S(x_in), and the problem formulation (p. 4) constrains total energy consumption per UAV, with the abstract (p. 1) proposing the FL framework within a leader-follower swarm while accounting for energy consumption. The claim's possibility modality ('can be further decreased') matches the source's framing; the only element our copy does not give is a quantified energy saving versus centralized offloading, and the 'edge AI' wording is the citing authors' framing (the source instead motivates distributed/on-device learning, citing wireless network intelligence at the edge [3]). (LLM, conf=0.82)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“The creation of channel models to represent such A2G links is challenging as many factors need to be considered, such as a UAV's movement/vibration, changes in the UAVs altitude, type of UAV, type of propagation environment, antenna movements and shadowing caused by the UAVs own frame, to name a few [115].”
§ A Survey on the Convergence of Edge Computing and AI for UAVs: Opportunities and Challenges > a) A2G communication:
What the sources say
- Supportedcites [115] source
Passage [1] (A. Air-to-Ground Channel Modeling, p. 9) states each factor the claim attributes to the source: UAV movement or vibration affecting channel characteristics, dependence on altitude and type of UAV, elevation angle and type of propagation environment, plus antennas' movements and shadowing caused by the UAV's body, concluding that 'capturing such factors is challenging in A2G channel modeling.' The citing sentence is a faithful paraphrase of that passage, with no qualifier dropped or added. (LLM, conf=0.95)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“The use of AI is not compulsory in computer vision techniques such as how Petricca et al. [103] can perform rust detection based on the number of pixels containing certain red components.”
§ A Survey on the Convergence of Edge Computing and AI for UAVs: Opportunities and Challenges > b) Real-time embedded computer vision processing:
What the sources say
- Supportedcites [103] source
The Abstract of Petricca et al. (2016) states that "For the classic approach, a classification based on the number of pixels containing specific red components has been utilized," and Section 2.1 (Classic Computer Vision Technique) details the OpenCV/HSV red-component masking, converting the mask to black-and-white and classifying any image with more than 0.3% white pixels as "rust" — no AI involved. Section 2 (Approach Used) frames the work as implementing "one version of classic computer vision (based on red component) and one deep learning model" for comparison, and the Discussion reports the non-AI OpenCV model reaching 69% total accuracy, so the cited work does demonstrate that AI is not compulsory for such computer-vision rust detection. The exemplifying clause the citation is attached to matches the source exactly. (LLM, conf=0.90)
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Despite current advances, UAVs still face challenges in performing tasks that are computationally intensive because of a UAV's limited processing power and battery lifetime [121].”
§ A Survey on the Convergence of Edge Computing and AI for UAVs: Opportunities and Challenges > b) A2A communication:
What the sources say
- Supportedcites [121] source
The Introduction (p. 1, Section 1) of the iTOA paper states almost verbatim that 'even with current advances, UAVs are faced with challenges in performing computationally intensive tasks due to limited battery lifetime and processing power', and reiterates that 'limited computing power and the energy supply of UAVs make the problem difficult and challenging'. Passage [4] (Section 1) likewise frames MEC offloading as a response to 'the challenges of the limited processing power and battery lifetime on UAVs'. The paraphrase carries all qualifiers of the source statement. (LLM, conf=0.95)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
62 more claims · 4 with their cited sources checked
Create a free account to see all 163 claims and to verify the 11 references still open.
- It may state two separate claims as one. 58
- Probably not a substantial claim. 53
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“International Data Corporation (IDC) predicts that by 2025, there will be 150 billion intelligent edge devices in use [72].”
From Edge AI: A survey (2023) · cites this paper as [72]
Experimental — accuracy not yet measured
Citations
Cites
LeCun, Y., Bengio, Y., & Hinton, G. (2015) · Nature · doi:10.1038/nature14539 · Cited by 84,126
Reinforcement Learning: An Introduction
Johnson, J. D., Li, J., & Chen, Z. (2000) · Neurocomputing · doi:10.1016/s0925-2312(00)00324-6 · Cited by 8,414
Wireless communications with unmanned aerial vehicles: opportunities and challenges
Zeng, Y., Zhang, R., & Lim, T. J. (2016) · IEEE Communications Magazine · doi:10.1109/mcom.2016.7470933 · Cited by 4,048
Unmanned Aerial Vehicles (UAVs): A Survey on Civil Applications and Key Research Challenges
Shakhatreh, H., Sawalmeh, A. H., Al-Fuqaha, A., Dou, Z., Almaita, E., Khalil, I., Othman, N. S., Khreishah, A., & Guizani, M. (2019) · IEEE Access · doi:10.1109/access.2019.2909530 · Cited by 2,529 · cited 10×
“For example, paper [11] in the first row considers the areas “UAV,” “UAV Technical Challenges,” and “UAV Applications” in high detail (dedicated sections or substantial subsections focused on the areas), considers the area “AI/ML” partially (mentioned in various parts of this article but in no…”
BackgroundInfluentialEdge Intelligence: Paving the Last Mile of Artificial Intelligence With Edge Computing
Zhou, Z., Chen, X., Li, E., Zeng, L., Luo, K., & Zhang, J. (2019) · Proceedings of the IEEE · doi:10.1109/jproc.2019.2918951 · Cited by 2,455 · cited 7×
“…and distributed security mechanism designs can be designed to establish user authentication and access control, model/data integrity, and mutual platform verification [8]. d) The use of federated learning, blockchain, or the integration of both (e.g., [37]) can further improve security/privacy.”
BackgroundMethodInfluentialSurvey of Important Issues in UAV Communication Networks
Gupta, L., Jain, R., & Vaszkun, G. (2016) · IEEE Communications Surveys & Tutorials · doi:10.1109/comst.2015.2495297 · Cited by 2,343
Precision Agriculture and Food Security
Gebbers, R., & Adamchuk, V. I. (2010) · Science · doi:10.1126/science.1183899 · Cited by 1,595
A survey on security and privacy of federated learning
Mothukuri, V., Parizi, R. M., Pouriyeh, S., Huang, Y., Dehghantanha, A., & Srivastava, G. (2021) · Future Generation Computer Systems · doi:10.1016/j.future.2020.10.007 · Cited by 1,480
Survey on Unmanned Aerial Vehicle Networks for Civil Applications: A Communications Viewpoint
Hayat, S., Yanmaz, E., & Muzaffar, R. (2016) · IEEE Communications Surveys & Tutorials · doi:10.1109/comst.2016.2560343 · Cited by 1,386
UAV Communications for 5G and Beyond: Recent Advances and Future Trends
Li, B., Fei, Z., & Zhang, Y. (2019) · IEEE Internet of Things Journal · doi:10.1109/jiot.2018.2887086 · Cited by 1,377
Research works printed but not linked: 19
- 2021, Comparing fog computing with edge computing
- Hernández, 2020, The role of AI in drones and autonomous flight, datascience aero blog
- Chao, 2020, A survey of UAV-based edge intelligent computing, Chin. J. Intell. Sci. Technol., 2, 227
- 2021, Global Anti-Drone Markets 2021–2026—Increasing Terrorism and Illicit Activities Through Drones & Rising Investment by Governments
- Hsieh, Gaia: Geo-distributed machine learning approaching {LAN$} speeds, 629
- Cepero, 2019, 4 Benefits of Edge Computing
- LaBrie, 2019, Top 5 Benefits of Edge Computing
- Smith, 2018, Understanding the pros and cons of cloud vs on-device AI FunkyKi
- Hiter, 2021, The Pros and Cons of edge computing, datamation
- 2020, New AI Solution to Shield Drones From Advanced Cyber Attacks
- Brown, 2019, 6 advantages of edge computing in 2020 and beyond, Toolbox
- 2018, What is deep learning and why you need it?
- Chawla, 2021, UAV delivery system, Int. J. Sci. Res. Eng. Manage., 5, 1
- Alosious, 2021, The how and why of using drones for search and rescue operations?
- Joern, 2015, Examining the use of unmanned aerial systems and thermal infrared imaging for search and rescue efforts beneath snowpack, Geomat. Nat. Hazards Risk, 8, 18
- Poljak, 2021, What Is Drone Light Show? Drone Display
- Zhan, 2020, 3,051 Drones Create Spectacular Record-Breaking Light Show in China
- Thorjussen, 2021, Everything You Ever Wanted to Know About Drone Light Shows
- 2021, Commercial Drone Market Size, Share & Trends Analysis Report by Product (Fixed-Wing, Rotary Blade, Hybrid), by Application, by End-Use, by Region, and Segment Forecasts, 2021–2028 Report Summary
Cited by
Singh, R., & Gill, S. S. (2023) · Internet of Things and Cyber-Physical Systems · doi:10.1016/j.iotcps.2023.02.004 · Cited by 572
Communication and Control in Collaborative UAVs: Recent Advances and Future Trends
Javaid, S., Saeed, N., Qadir, Z., Fahim, H., He, B., Song, H., & Bilal, M. (2023) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2023.3248841 · Cited by 345
AI for UAV-Assisted IoT Applications: A Comprehensive Review
Cheng, N., Wu, S., Wang, X., Yin, Z., Li, C., Chen, W., & Chen, F. (2023) · IEEE Internet of Things Journal · doi:10.1109/jiot.2023.3268316 · Cited by 307
Edge Computing on IoT for Machine Signal Processing and Fault Diagnosis: A Review
Lu, S., Lu, J., An, K., Wang, X., & He, Q. (2023) · IEEE Internet of Things Journal · doi:10.1109/jiot.2023.3239944 · Cited by 295
Chataut, R., Phoummalayvane, A., & Akl, R. (2023) · Sensors · doi:10.3390/s23167194 · Cited by 329
Metadata
- Authors
- McEnroe, P., Wang, S., & Liyanage, M.
- Affiliations
- University College Dublin (IE)
- Venue
- IEEE Internet of Things Journal
- Published
- 2022-09-01
- Volume
- 9
- Issue
- 17
- Pages
- 15435-15459
- ISSN
- 2327-4662 (ISSN-L)
- 2372-2541
- OpenAlex
- W4285105424
- Semantic Scholar
- 0c6c9c901d9c10639a12067804dc90c282fc0b92
- Cited by
- 482
- References
- 163
- Citations by year
- 2026: 96 · 2025: 145 · 2024: 142 · 2023: 85 · 2022: 14
- FWCI
- 42.79
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Locations
- IEEE Internet of Things Journal (publishedVersion, cc-by)
- Research Repository UCD (University College Dublin) (submittedVersion, cc-by)
- Funders
- Science Foundation Ireland (SFI) Center for Research Training in Machine Learning (18/CRT/6183)
- SFI CONNECT Center Phase 2 (13/RC/2077_P2)
- 6Genesis Flagship Project (318927)
- Topics
- Physical Sciences › Computer Science › Computer Networks and Communications › IoT and Edge/Fog Computing
- Physical Sciences › Engineering › Aerospace Engineering › UAV Applications and Optimization
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Keywords
- edge computing, edge AI, unmanned aerial vehicles, autonomous navigation, formation control, power management, computer vision, low-latency decision-making, energy efficiency, 5G, drone delivery, civil infrastructure inspection, precision agriculture, search and rescue
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: A Survey on the Convergence of Edge Computing and AI for UAVs: Opportunities and Challenges. https://app.cochat.ai/papers/10.1109/jiot.2022.3176400
- The paper
- McEnroe, P., Wang, S., & Liyanage, M. (2022). A Survey on the Convergence of Edge Computing and AI for UAVs: Opportunities and Challenges. IEEE Internet of Things Journal. https://doi.org/10.1109/jiot.2022.3176400
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.