State-of-the-Art Review on the Application of Unmanned Aerial Vehicles (UAVs) in Power Line Inspections: Current Innovations, Trends, and Future Prospects

Abstract

Unmanned aerial vehicles (UAVs) make power line inspections more safe, efficient, and cost-effective, replacing risky manual checks and expensive helicopter surveys while overcoming challenges like stability and regulations. The aim of this study is to conduct a systematic review of the application of UAVs for power line inspections. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) methodology is implemented to ensure a structured and comprehensive review process. The Scopus database is used to identify relevant publications, and after screening and applying eligibility criteria, 75 documents were selected for further analysis. The study results show a shift toward predictive maintenance, multi-UAV operations, and real-time data analysis. However, challenges remain, including UAV–grid connectivity, resilience to extreme weather, and large-scale automation. This work provides key insights into technological and algorithmic advancements and research trends on UAV-based power line inspections while pointing out gaps in the existing literature. Finally, future research directions to advance UAV-based power line inspections are suggested.

1. Introduction

UAVs, commonly known as drones, are aircraft that operate without an onboard human pilot [1]. They can be controlled remotely or autonomously [2]; they have gained significant importance across various sectors due to their versatility and cost-effectiveness [3,4,5] and can be used for various purposes, including military operations, surveillance, and civilian applications [6,7]. In the previous years, their global market grew from USD 3.4 billion in 2008 to USD 7.3 billion in 2017 [8], driven by applications like real-time monitoring and precision agriculture [9,10]. In the coming years, the global UAV market is projected to experience significant growth. By 2025, the market capacity is expected to triple [11], with estimates ranging from USD 45 billion for civil infrastructure applications [9] to USD 125 billion for civilian commercial drones by 2032 [12]. The USA, China, and France are the leading producers of commercial drones [11]. Key sectors driving market growth include agriculture, pipeline and railroad inspection, construction, and maritime applications [13].

UAVs come in different types, such as (1) fixed-wing hybrid, (2) fixed-wing, (3) single-rotor, and multirotor [14], with fixed-wing and multirotor being the most popular [15]. A fixed-wing hybrid combines the characteristics of fixed-wing UAVs and quadcopters, merging the advantages of both designs for military and civil applications and addressing the limitations inherent in fixed-wing and rotary-wing drones [16]. Fixed-wing UAVs are widely used in various sectors, including agriculture for monitoring, mapping, and dropping, as well as in real estate, film, TV, oil and gas, construction, fisheries, wildlife surveillance, water management, and security [17,18,19,20,21,22]. They are particularly effective in surveillance and reconnaissance during disasters and for monitoring overhead transmission lines, offering extensive imagery coverage and real-time low-latency monitoring [23,24,25,26].

Single-rotor UAVs are noted for their versatility and efficiency, especially in agriculture for spraying fertilizers and pesticides [27]. Despite challenges in flight stabilization and control systems, they have significant potential in various applications [28,29,30,31,32]. Multirotor UAVs, comprising bicopters, tricopters, quadcopters, hexacopters, and octocopters, are highly versatile and used in military operations, delivery, rescue, weather forecasting, agriculture, entertainment, communication, and defense [33,34,35,36,37]. They also play roles in disaster management, construction safety, precision agriculture, and power grid supervision, with ongoing advancements in power sources, energy management, obstacle avoidance, and human-supervised inspection systems [38,39].

Table 1. Different classifications of UAV types and their areas of application.

Ref. No.	UAV Type	Area of Application
[16]	Fixed-wing hybrid	Military and civil
[18,19,20,21,22,23]	Fixed-wing	Monitoring, mapping, agriculture, real estate, film and TV, oil and gas, construction, fisheries, wildlife surveillance, water management, security surveillance, reconnaissance in disaster, and power line inspections
[24,25,26,27,28,29]	Single-rotor	Agriculture, particularly for spraying fertilizers and pesticides
[31,32,33,34,35]	Multirotor	Military, delivery and shipping, rescue, search operations, forecasting weather, agriculture, entertainment, communication, video surveillance, defense, disaster management, surveillance, and construction industry

presents different classifications of UAV types and their areas of application.

Traditionally, inspecting power lines has been a labor-intensive foot patrol and hazardous task, requiring workers to scale heights and navigate difficult terrains and costly manned helicopter methods [40,41,42]. Foot patrols [43,44] are slow and often miss defects, while manned helicopters, despite being equipped with advanced cameras, are unsafe and expensive, as noted in [45,46]. Meanwhile, [47,48] criticized these outdated methods for their inefficiency and high risk. Maintenance under high voltage conditions requires extreme labor and poses significant risks, as highlighted in [49,50,51]. Conversely, power lines will forever be exposed to risks such as fires (i.e., bushfires or wildfires), hostile climatical environments (i.e., snow, strong winds, rain, sunlight, extreme temperatures), vegetation interferes, bird nests, extra high voltages, mechanical tension, etc. [52,53,54,55], and these phenomena sometimes result in physical defects and power interruption. The economic impact of power interruptions is significant, with annual costs estimated at USD 150 billion [56], while the electricity consumer-related cost is estimated at USD 80 billion annually, with a potential range of USD 30–130 billion [57], as observed in the U.S. Considering these factors, adopting advanced line inspection methods like UAVs is crucial for improving safety, accuracy, and cost-effectiveness [58]; also, navigation techniques are being explored to enhance inspection capabilities and accuracy [59].

UAVs are increasingly important for power line inspections due to their ability to improve efficiency, safety, and cost-effectiveness compared to traditional methods [60,61,62]. These systems employ various sensors, including LiDAR, optical cameras, and thermal imaging, to detect faults and collect comprehensive data on electrical infrastructure [63,64,65]. Advanced path planning algorithms and control strategies, such as Particle Swarm Optimization and fuzzy PD control, enable UAVs to navigate complex environments and maintain optimal inspection distances [61,66], while vision-based detection systems enhance UAV navigation and power line identification [67,68]. Despite challenges such as flight stability, safety concerns, environmental factors, sensor integration, and regulatory issues [69], UAVs offer significant advantages over traditional methods, such as reduced costs, improved safety, enhanced efficiency, and infrastructure disruption [70,71,72]. Therefore, as UAV inspection technology matures, efforts are being made to standardize practices and promote international adoption, particularly in emerging countries with developing power networks [73].

Several key studies have laid the groundwork in this field; for example, Alhassan et al. (2020) [74] and Yang et al. (2020) [75] explored key challenges and methodologies, while Mohsan et al. (2022) [4] addressed UAV security concerns. Research in 2023 [76] further investigated UAV swarms and infrastructure needs. Chamola et al. (2021) [77] and Foudeh et al. (2021) [78] examined vulnerabilities and fault detection. Williams et al. (2001) [79], Martins et al. (2020) [80], and Surantha et al. (2022) [81] highlighted UAV benefits, such as safety and energy efficiency, while Liu et al. (2020) [82,83] and Sun et al. (2021) [84] advanced autonomous inspection techniques. However, gaps remain in regulatory issues, operational limitations, and multi-platform integration. For instance, Nguyen et al. (2018) [85] and Gonçalves et al. (2023) [86] focused on vision-based methods but lacked broader application contexts, while the security was addressed without tackling real-world challenges. Fault detection and path planning have improved [60,67], but human acceptance and reliability need more focus [71,87]. Wang et al. (2021) [88] discussed historical UAV use, Sun et al. (2021) [84] examined navigation, and Fahmani et al. (2022) [89] proposed safe distance estimation.

While the existing literature on UAVs for overhead power line inspections covers a broad range of topics, there is a notable gap in the literature concerning a comprehensive review analysis of fault identification strategies, algorithms and UAV-based systems, challenges and developments, robotic vision and UAV innovations, UAV technologies for autonomous operations, and AI-driven UAV systems. Therefore, this study aims to conduct a systematic review of the application of UAVs for power line inspections with a special focus on innovative advancements, related trends analysis, and future research prospects. The key study objectives include the following:

• Adopt the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework to ensure a systematic, transparent, and comprehensive review process.
• Review advancements in UAV-based power line inspections, including fault detection, advanced algorithms, system challenges, robotic vision, autonomous UAV technologies, and AI-driven inspection systems.
• Highlight the emerging trends and observations in the UAV-based power line inspection research area from 2019 to 2023.
• Provide recommendations for future research directions based on the gaps identified in the current literature.

This work highlights key technological and algorithmic advancements, analyzes research trends, points out gaps in the existing literature through discussions and observations, and suggests future research directions to improve UAV-based power line inspections, providing valuable insights into the field.

This paper is organized into the following sections: Section 2 covers the progressions of UAVs in power system inspection. Section 3 outlines the research methods applied during this study. Section 4 discusses the results obtained and provides an analysis of the findings. Section 5 offers a summary of the overall study discussions and observations. Finally, Section 6 presents the conclusions drawn from this study and suggests potential areas for future research.

2. Progressions of UAVs in Power System Inspection

The use of UAVs in power system inspections has progressed rapidly, with various models differing in size, range, and endurance [90]. Drones have become crucial in geospatial applications [91], and the global UAV market is valued at over USD 127 billion, with USD 9 billion in 2020 dedicated to inspection technologies [92]. In the U.S., the UAV market grew from USD 40 million in 2012 to USD 1 billion in 2017 and is forecasted to contribute USD 31–USD 46 billion to the GDP by 2026 [93]. Recent studies highlight improvements in inspection systems through machine learning [84], UAV performance assessments [94], and autonomous inspection systems. Advanced systems for power line inspections were also developed [95,96], while in [97], UAV use in utility inspections was documented, showcasing significant progress in this area. These studies collectively demonstrate the significant progress in the use of UAVs for power system inspection, with a focus on the developments around autonomous inspection, efficiency, and accuracy. Figure 1 is a conceptual representation of an autonomous power line inspection system, showcasing UAV-based monitoring, cloud-based data processing, and automated battery replacement.

The UAV task begins with an upload via a signal tower, allowing it to autonomously inspect power insulators and bird nests. Photos are uploaded to a cloud-based system for analysis, and the inspection report is stored in a database. The UAV returns to a base station for battery replacement. Accurate inspections rely on equations for flight height and tower identification [89], as demonstrated in Equations (1)–(3) below. Equation (1) can be used to calculate the appropriate height $h$ at which the UAV should fly to maintain an optimal view of the power line, ensuring that the UAV captures the necessary details while maintaining a safe distance from the line, while the additional factor $\mathrm{\alpha }$ might include safety margins or adjustments for specific operational conditions:

$$\mathrm{h}=\mathrm{t}\mathrm{a}\mathrm{n}\left({\displaystyle \frac{{\mathrm{F}\mathrm{O}\mathrm{V}}_{\mathrm{v}}}{2}}\right)\times \mathrm{d}+\mathrm{\alpha }$$

(1)

where h is the height that should be maintain by UAV above the power line, using the vertical field of view $\left({FOV}_v\right)$, the horizontal distance d, and an adjustment factor $\alpha$. Equation (2) allows the UAV to accurately calculate the direction of a transmission tower with respect to true north. The first term accounts for the UAV’s rotation, and the second term adjusts this based on the position of the tower within the camera’s field of view. This calculation is essential for precise navigation and mapping during power line inspections and can be expressed as follows:

$$\mathrm{\theta }=\mathrm{\omega }\mathrm{t}+{\mathrm{t}\mathrm{a}\mathrm{n}}^{-1}\left(\mathrm{t}\mathrm{a}\mathrm{n}\left({\displaystyle \frac{{\mathrm{F}\mathrm{O}\mathrm{V}}_{\mathrm{h}}}{2}}\right)\left(1-{\displaystyle \frac{2{\mathrm{W}}_{\mathrm{T}}}{\mathrm{W}}}\right)\right)$$

(2)

where ω is the rotational speed (in degrees or radians per second), t is the rotational time duration (in seconds), $tan \left({FOV}_h/2\right)$ is the tangent of half of the horizontal field of view of the optical camera, and $(1-{2W}_T/W)$ is the relative horizontal position of the transmission tower within the image, where $W_T$ is the width coordinate value corresponding to the center position of the bounding box (the detected transmission tower) and W is the total width of the image. Finally, the inverse tangent function is used to convert the tangent value back to an angle. As explained above, the following equation evaluates the level of ascent δ for a transmission tower.

Accurately identifying and distinguishing transmission towers from other tall structures, like large trees, is crucial. One effective characteristic for this purpose is the steep ascent or vertical profile of the transmission tower. This characteristic can help avoid misrecognition because both transmission towers and large trees may exhibit high vertical density in their point cloud data. Thus, Equation (3) can be used to evaluate the level of ascent δ for a transmission tower.

$$\mathrm{\delta }={\displaystyle \frac{1}{8}}\sum _{\mathrm{i}=1}^8\left|\mathrm{H}-{\mathrm{H}}_{\mathrm{s},\mathrm{i}}\right|$$

(3)

where H represent the height of the transmission tower and $H_{s, i}$ are the heights of the surrounding points in the environment, where subscript s,i indicates that there are eight surrounding points considered in the calculation. The summation aggregates the elevation differences for all eight surrounding points, and the division by 8 provides a normalized measure of the ascent level.

The optimal flight altitude for UAVs is crucial, balancing image resolution and power consumption [99]. Studies have shown a correlation between flight altitude and defect detection in photovoltaic modules [100,101]. For power line inspections, maintaining a distance of 15–20 m from infrastructure while flying at least 5 m/s is recommended to meet safety and efficiency requirements [102]. UAVs equipped with various sensors, including visible light cameras, infrared thermal imaging, and GPS, can effectively detect faults and obstacles within power line corridors [103,104]. However, challenges remain in data processing, automation, and regulatory frameworks [105], highlighting the need for further research and development in this field.

3. Research Methodology

This study uses the Systematic Literature Review (SLR) methodology to examine UAV applications in power line inspections, focusing on structured literature synthesis [106]. This methodology is particularly important in fields with a large volume of literature, such as engineering, and it follows key phases: planning, conducting, and reporting [107,108] on this review. Systematic reviews remain underused in Library and Information Science [109], though statistical methods help address publication bias and nonrandomized evidence synthesis [110]. Study quality and interpretation are critical [111]. To ensure consistency, this research follows the PRISMA guidelines [112].

3.1. Identification

The initial phase of this study involved a comprehensive search for relevant documents within the Scopus database. This process yielded a total of 238 initial records. The search was conducted using a strategic combination of keywords to ensure thorough coverage of the topic. Specifically, the keywords “UAV”, “unmanned aerial vehicle”, and “drone” were used in conjunction with “power line inspections”. This approach was designed to capture a wide range of the literature pertaining to the use of unmanned aerial vehicles in the context of power line inspection operations. The resulting dataset forms the basis for subsequent analysis and review.

3.2. Screening

The filtering process was initiated to refine the initial set of 238 documents. The first criterion applied was the range of publication years, limited to the period from 2019 to 2023, which resulted in 147 remaining documents. Next, the search was restricted to the subject area of “engineering”, further narrowing the pool to 99 documents. Subsequently, the document type was filtered to include only articles and conference papers, reducing the count to 94. Finally, the language criterion was applied, selecting only English-language documents, which left 92 documents. These 92 documents were then exported for further analysis and qualified for the subsequent screening process.

3.3. Eligibility

Table 2. Emerging technologies, learning algorithms, and associated developments in autonomous systems and data processing.

Ref. No.	Year	Authors	Ref. No.	Year	Authors
4.2 Innovative Advancements in UAV-based Power Line Inspections
4.2.1 UAV-Based Approaches and Algorithms for Power Line Fault Detection and Inspection
[113]	2019	Zhang et al.	[114]	2020	Wang et al.
[115]	2019	Azevedo et al.	[116]	2020	Fang et al.
[117]	2021	Solilo et al.	[118]	2019	Xu et al.
[119]	2022	Li et al.	[120]	2022	Akhmetov et al.
[121]	2019	Liu and Zhong	[122]	2019	Korki et al.
[123]	2022	Dai et al.	[124]	2020	Fang et al.
[125]	2023	Han et al.	[126]	2022	Zhong et al.
[127]	2023	Xing et al.	[128]	2023	Shuang et al.
[129]	2023	Li et al.	[130]	2023	Hoang and Ebeid
[131]	2021	Ding et al.	[132]	2021	Wang et al.
[133]	2021	Demkiv et al.	[134]	2021	Zheng et al.
[135]	2020	Siddiqui and Park	[136]	2023	Li et al.
[137]	2020	Vemula and Frye	[138]	2023	Liang et al.
[139]	2022	Song et al.	[140]	2019	Yang et al.
[141]	2019	Pienroj et al.	[142]	2021	Ayoub and Schneider-Kamp
[143]	2023	He et al.	[144]	2022	Han et al.
[145]	2023	Zheng et al.	[146]	2019	Hamelin et al.
[147]	2022	Shen et al.	[148]	2023	Kyuroson et al.
[149]	2019	Titov et al.	[150]	2023	An et al.
[151]	2023	Wang et al.
4.2.2 Challenges and Developments
[152]	2021	Silano et al.	[153]	2023	Rad et al.
[154]	2022	Mostafa et al.	[155]	2023	Rad et al.
[156]	2022	Duan et al.	[157]	2022	Skriver et al.
[158]	2022	Cesoni et al.
4.2.3 Robotic Vision and UAV Innovations
[159]	2023	Ye et al.	[160]	2023	Wang et al.
[161]	2022	Wang et al.	[162]	2019	Wang et al.
[163]	2023	Damigos et al.	[164]	2020	Cantieri et al.
[165]	2022	Liu et al.	[166]	2021	Sun et al.
[167]	2020	Pan et al.	[168]	2022	Calvo et al.
[169]	2021	Silano et al.
4.2.4 UAV Technologies for Autonomous
[170]	2021	Wang	[171]	2021	Li et al.
[172]	2021	Savva et al.	[173]	2019	Takaya et al.
[174]	2020	Schofield et al.	[175]	2021	Iversen et al.
[176]	2021	Lopez et al.	[177]	2023	Alexiou et al.
[178]	2020	Perez-Jimenez et al.	[179]	2023	Zhao et al.
[180]	2020	Cantieri et al.
4.2.5 AI-Driven UAV Systems
[181]	2022	Moreno et al.	[182]	2022	Rangel et al.
[183]	2023	Tsellou et al.	[184]	2022	Du et al.
[185]	2021	Liu et al.	[186]	2020	Medrano et al.
[187]	2019	Constantin and Dinculescu

lists 75 research papers published between 2019 and 2023, covering technological advancements, learning algorithms, and autonomous systems. Initially, 92 papers were selected for review, but some had to be excluded. Of the excluded papers, 10 were inaccessible, 5 were review articles that did not meet the study criteria, and 2 were unrelated to UAVs and power line inspections. After filtering out these papers, 75 remained, forming the core dataset for further analysis.

3.4. Classification and Synthesis of Key Themes

The synthesis of the 75 finalist documents involved a comprehensive thematic classification, grouping the research findings into key themes, such as detection and fault identification strategies, which examined methods for identifying abnormalities and faults in power lines; challenges and developments, which focused on advancements and barriers in UAV technology; robotic vision and UAV innovations, which explored improvements in robotic vision used for inspections; UAV technologies for autonomous operations, which highlighted progress in UAV autonomy and navigation; advanced algorithms and UAV-based systems, which showcased algorithmic improvements that enhance UAV performance; and, finally, AI-driven UAV systems, which investigated how AI is being used to boost decision-making and fault diagnosis. This classification offered a clear analysis of how UAV technology is evolving in the context of power line inspections and related applications.

4. Results and Discussion

4.1. Bibliometric Analysis

4.1.1. Annual Scientific Production and Application Areas of UAV Usage in Power Line Inspections

Figure 2a presents the trend in annual and cumulative publications from 2019 to 2023. In 2019, there were 12 articles, but this number slightly dropped to 11 in 2020. From 2021 onward, the number of publications gradually increased, reaching 15 in 2021, 17 in 2022, and 20 in 2023. As a result, the cumulative total grew from 12 in 2019 to 75 by 2023. The data show a steady rise in research output, with the most noticeable increase occurring in the final two years, indicating a growing focus on this area of study. Figure 2b shows how the research publications are distributed across different themes.

The largest share, with 21 articles, focuses on detection and fault identification strategies, indicating a strong interest in this area. Algorithms and UAV-based systems follow closely with 18 articles, reflecting efforts to integrate UAVs with algorithmic solutions. Robotic vision and UAV innovations and UAV technologies for autonomous operations each account for 11 articles, highlighting ongoing research into UAV applications for inspection and automation. Meanwhile, challenges and developments and AI-driven UAV systems each have seven publications, showing that while these topics are important, they have received comparatively less attention. Overall, the distribution suggests a growing focus on UAVs and advanced algorithms for detection, fault identification, and automation.

4.1.2. Most Cited Journals and Conferences in UAV Power Line Inspections

Table 3. Top 10 most cited journals and conference proceedings in the field.

Journal and Conference Titles	Total Citations
Sensors	134
Energies	35
IEEE International Conference on Intelligent Robots and Systems	35
Electronics	63
IEEE Robotics and Automation Letters	31
2019 23rd International Conference on System Theory, Control and Computing, ICSTCC 2019—Proceedings	27
2019 International Conference on Unmanned Aircraft Systems, ICUAS 2019	25
AIAA/IEEE Digital Avionics Systems Conference—Proceedings	25
Multimedia Tools and Applications	20
IEEE Transactions on Reliability	20

highlights the top 10 most cited journals and conferences in the field, showing where key research is concentrated. Sensors stands out with the highest citations (134), indicating its strong influence. Electronics (63) and IEEE International Conference on Intelligent Robots and Systems (35) also have significant impact. The list includes both high-impact journals and specialized conferences, reflecting diverse research interests, particularly in automation, robotics, and system control.

4.1.3. Top 10 Most Cited Publications in UAV Power Line Inspections

Table 4. Top 10 most cited publications in the field.

Authors	Title	Year	Cited by
Azevedo et al. [115]	LiDAR-based real-time detection and modeling of power lines for unmanned aerial vehicles	2019	56
Siddiqui and Park [135]	A Drone Based Transmission Line Components Inspection System with Deep Learning Technique	2020	35
Iversen et al. [175]	Design, Integration and Implementation of an Intelligent and Self-recharging Drone System for Autonomous Power line Inspection	2021	35
Sun et al. [166]	An industrial quadrotor uav control method based on fuzzy adaptive linear active disturbance rejection control	2021	32
Silano et al. [152]	Power Line Inspection Tasks with Multi-Aerial Robot Systems Via Signal Temporal Logic Specifications	2021	31
Takaya et al. [173]	Development of UAV system for autonomous power line inspection	2019	27
Ayoub and Schneider-Kamp [142]	Real-time on-board deep learning fault detection for autonomous UAV inspections	2021	26
Hamelin et al. [146]	Discrete-time control of LineDrone: An assisted tracking and landing UAV for live power line inspection and maintenance	2019	25
Vemula and Frye [137]	Mask R-CNN powerline detector: A deep learning approach with applications to a UAV	2020	25
Cantieri et al. [180]	Cooperative uav–ugv autonomous power pylon inspection: An investigation of cooperative outdoor vehicle positioning architecture	2020	23

presents the top 10 most cited publications in UAV power line inspections, highlighting key research contributions in the field. The most cited paper, “LiDAR-based real-time detection and modeling of power lines for unmanned aerial vehicles” by [115], received 56 citations, reflecting the growing importance of LiDAR technology in UAV-based inspections. Other highly cited works focus on deep learning applications, autonomous UAV systems, and cooperative aerial–ground inspection strategies. The presence of multiple studies from 2019 to 2021 indicates a rapidly evolving research landscape, with increasing emphasis on AI-driven fault detection, multi-robot collaboration, and real-time data processing for power line monitoring.

4.2. Innovative Advancements in UAV-Based Power Line Inspections

4.2.1. UAV-Based Approaches and Algorithms for Power Line Fault Detection and Inspection

Object detection plays a pivotal role in identifying bird nests and insulators, enhancing power line detection performance. For instance, in [113,114], Zhang et al. (2019) and Wang et al. (2020) introduced CNN-based frameworks for automated detection of abnormalities in power line infrastructure, while Azevedo et al. (2019) proposed PL2DM [115], which leverages LiDAR for real-time detection of power lines even in challenging environmental conditions. Image processing and recognition techniques [116,117], as outlined by Fang et al. (2020) and Solilo et al. (2021), have been employed for detecting transmission line defects and tracking power lines using color transformation techniques, respectively.

Innovative control strategies in support of these methods, such as the weighted consensus control method introduced by Xu et al. (2019), ensure robust fault detection and global load balancing in multi-drone systems utilized for power line inspection [118]. Additionally [119,120], advancements in object detection, demonstrated by Li et al. (2022) and Akhmetov et al. (2022), have enabled the real-time detection of bird nests and insulators using optimized YOLO (You Only Look Once) variants. In ensuring collaborative model training for insulator detection while preserving data privacy, Liu and Zhong (2022) proposed the integration of federated learning [121]. Korki et al. (2019) introduced the concept designs for automatic fault detection [122], underlining the pivotal role of UAVs in this field. Building upon this, Dai et al. (2022) proposed CODNet, a state-of-the-art method made for UAV navigation [123], while in complementing these efforts [124], Fang et al. (2020) emphasized the importance of efficient detection techniques and precise tower positioning for fault identification in power line inspection. Moreover, in [125], Han et al. (2023) conducted research on real-time multi-scale fault detection algorithms, which are specifically designed for UAV inspection, showcasing enhanced efficiency and accuracy compared to previous models.

Further advancing the capabilities of UAV-based inspection [126], Zhong et al. (2022) discussed a spatio-temporal correlation-based method for anomaly detection and recovery prediction in UAV flight data, while Xing et al. (2023) introduced an autonomous inspection system employing Model Predictive Control (MPC) and learning-based detectors for optimized line tracking and collision avoidance [127], further enriching the capabilities of UAV-based power line inspection. Additionally [128], Shuang et al. (2023) developed the RSIn-Dataset and proposed the YOLOv4++ network for insulator detection, further enriching the repertoire of detection techniques in UAV-based power line inspection. Li et al. (2023), in [129], aimed to improve power line detection performance in adverse weather conditions using the CSUNet image restoration model and a large-scale PLCD dataset, demonstrating enhanced object detection results. Hoang and Ebeid (2023) [130] and Ding et al. (2021) [131] established UAV-centric approaches for overhead transmission line inspections. By integrating sensor fusion techniques and Light Detection and Ranging (LiDAR) data analysis, navigation precision and vegetation encroachment detection were refined and shown to be vital for ensuring infrastructure integrity. Furthermore, Wang et al. (2021) [132] studied federated learning methods to address privacy concerns in collaborative model training, and this ensured data security while facilitating reliability improvement in power infrastructure maintenance. Demkiv et al. (2021), in [133], presented a stereo thermal vision system for Micro-Unmanned Aerial Vehicles (MUAVs) to inspect electrical power lines, detecting overheated equipment to prevent outages.

Enhancing defect detection precision through advanced algorithms is key to minimizing downtime and operational costs. For example, the works of Zheng et al. (2021) [134] and Siddiqui and Park (2020) [135] both delved into real-time object detection and defect identification within power line inspection systems. Leveraging convolutional neural networks (CNNs) and recurrent neural networks (RNNs), they refined methodologies for precise component identification and anomaly detection. Building on this, Li et al. (2023) [136] and Vemula and Frye (2020) [137] further optimized inspection processes with automated systems driven by deep learning methodologies, collectively aiming to streamline inspection workflows and bolster infrastructure reliability.

Liang et al. (2021) [138] and Song et al. (2022) [139] made significant strides in enhancing power line image quality and extraction through GAN-based deblurring frameworks and Mask RCNN-based algorithms, respectively, improving the reliability of inspection imagery and accurate power line identification. Yang et al. (2019) [140] employed deep learning methodologies for automatic insulator self-shattering identification, aligning with the overarching goal of enhancing defect detection precision. Pienroj et al. (2019) [141], Ayoub and Schneider-Kamp (2021) [142], and He et al. (2023) [143] revolutionized UAV-based inspection by deploying Deep Reinforcement Learning (DRL), real-time fault detection algorithms, and coordinated multi-UAV operations.

Han et al. (2022) [144], Zheng et al. (2023) [145], and Hamelin et al. (2019) [146] collectively explored various facets of UAV navigation during power line inspections, and their research encompassed power line segmentation optimization, reinforcement learning-based navigation strategies, and discrete-time control algorithms for trajectory planning and obstacle avoidance. These efforts aimed to enhance UAV autonomy and navigation efficiency, which are crucial for successful and safe inspection operations. Strategies like task offloading based on deep reinforcement learning, as proposed by Shen et al. (2022) [147], offered promising avenues for future research, ensuring enhanced safety, and efficiency in power infrastructure maintenance through the integration of cutting-edge algorithms and technologies in UAV-based power line inspection.

Kyuroson et al. (2023) [148] introduced unsupervised machine learning frameworks tailored for vegetation encroachment detection and precise power line structure analysis by employing probabilistic graphical models and clustering algorithms to enhance segmentation and classification tasks and further refining inspection methodologies and bolstering infrastructure resilience. Titov et al. (2019) [149] introduced a sophisticated deep learning-based system for utility pole detection and classification, aligning with An et al.’s (2023) [150] WaveGNet proposal for continuous refinement of power line detection methodologies. A method for evaluating UAV inspection image quality focused on power line devices was proposed by Wang et al. (2023) in [151], utilizing DeepLabv3+ and DenseNet121, which can serve as a foundation for enhancing image processing algorithms in aerial inspection tasks.

4.2.2. Challenges and Developments in UAVs for Monitoring

Recent research has focused on optimizing UAV trajectories, addressing electromagnetic interference (EMI), and enhancing UAV design for harsh environments. The following reviews explore these innovations and their challenges. As discussed by Silano et al. (2021) in [152], the generation of trajectories with explicit time requirements and adherence to velocity and acceleration constraints provides a flexible and effective solution for inspection tasks. In [153], Rad et al. (2023) underlined the impact of magnetic fields on drone sensors, stressing the necessity of considering these effects in future drone navigation research. Further advancements in UAV design were discussed in [154] by Mostafa et al. (2022), tackling challenges related to flying near transmission lines, such as EMI conditions, obstacle avoidance systems, and the impact of EMI on Unmanned Aircraft System (UAS) sensors.

In addition to these advancements, in [155], Rad et al. (2023) presented a modular high-voltage transformer designed to study electric field impacts on drones, with experimental validation of its performance. Duan et al. (2022), in [156], introduced a novel UAV power patrol system based on wireless power transfer technology, addressing the limited application of UAVs in long-distance power line inspection. Furthermore, Skriver et al. (2022) [157] highlighted the experimental demonstration and evaluation of the HEIST methodology for recording EMI for UAVs approaching energized power lines. They emphasized the development of software robust toward EMI for safe UAV operation in harsh environments. Cesoni et al. (2022) [158] discussed the increasing use of UAVs for inspecting power lines and the challenges posed by the electromagnetic fields of high-voltage power lines.

4.2.3. Robotic Vision and UAV Innovations for Enhanced Power Line Inspections

Robotic vision techniques offer diverse approaches to improve the efficiency, accuracy, and safety of power line inspections while covering different aspects such as stabilization, image quality assessment, thermal detection, positioning accuracy, and real-time mapping. In [159], Ye et al. (2023) discussed the significance of large-scale flapping wing robots (FWR) with airborne vision, presenting a novel digital video stabilization method validated through experiments with a 2.2 m wingspan FWR, which could enhance the stability of footage captured during power line inspections. In another study [160], Wang et al. (2023) designed a flexible towed aerial robot system for stabilizing X-ray cameras in power line inspections, validated through real-world experiments. Wang et al. (2022) and Wang et al. (2019), in [161,162], introduced mechanical systems for UAV power line inspections, enhancing safety and efficiency. The 2022 study involved a UAV-mounted system tested in the field, while the 2019 study combined UAV and climbing robot technologies for improved maintenance. Damigos et al. (2023), in [163], revealed that 5G-enabled UAVs for sensor data throughput can contribute to real-time data processing and communication in various robotic systems, enhancing their autonomy and responsiveness.

As stated by Cantieri et al. (2020) in [164], the exploration of augmented reality (AR) tags for UAV positioning accuracy can be integrated into robotic navigation systems for precise localization and mapping in dynamic environments. In addition, Liu et al. (2022), in [165], proposed a PTI-SLAM framework that can be applied to power tower inspection tasks, improving the efficiency and accuracy of robotic navigation and mapping in challenging environments. The Fuzzy-LADRC control method for quadrotor UAVs was introduced by Sun et al. (2021) in [166], which can be utilized to enhance the stability and flexibility of aerial robotic platforms in various missions. To investigate the secrecy outage performance of UAV systems, Pan et al. (2020), in [167], derived analytical expressions for secrecy outage probability and validating through Monte Carlo simulations, as this can also inform the development of secure communication protocols for robotic networks and ensure data integrity. Calvo et al. (2022), in [168], proposed software architecture leveraging a Robot Operating System (ROS) to coordinate heterogeneous aerial vehicles for power line inspections and maintenance tasks. This architecture integrated a High-Level Planner and Agent Behavior Manager on each UAV while facilitating seamless communication and coordination among the team. Silano et al. (2021), in [169], emphasized the urgent need for safe and efficient maintenance and inspection procedures in electrical power lines, particularly focusing on robotic solutions by proposing a software architecture that is based on ensuring compliance with safety requirements and mission success.

4.2.4. Advancements in UAV Technology for Autonomous Power Line Inspections

As highlighted by Wang (2021) in [170], the advancements in autonomous navigation, visual guidance, and robust algorithms made for accurate measurements under diverse lighting conditions. Li et al. (2021), in [171], complemented this by presenting a mathematical model grounded on electric field distribution around transmission lines, ensuring resilience to environmental variables and real-time operational efficacy, substantiated through rigorous experimental validation. The introduction of ICARUS as an autonomous inspection platform integrating UAVs and AI for power infrastructure scrutiny, as articulated by Savva et al. (2021) [172], signified a pivotal hybrid navigation strategy augmented by a suite of sensors for dependable data acquisition. Takaya et al. (2019), in [173], expanded on a UAV system engineered for the autonomous inspection of energy networks, detailing its hardware configuration and software control systems validated through careful flight experimentation.

In [174], Schofield et al. (2020) presented a detailed framework for autonomous power line inspection, addressing key components and future developmental trajectories, while Iversen et al. (2021), in [175], explained an integrated drone system made for autonomous power line inspection, showcasing its ability in energy harvesting, AI-driven fault detection, and successful mission execution in diverse weather conditions. The autonomous UAV system presented by Lopez et al. (2021) in [176], for insulator cleaning on power lines, expressed a transformative approach toward mitigating risks and reducing operational costs. In support of this innovation, Alexiou et al. (2023), in [177], introduced a visual-guided navigation model stressing stability, robustness, and real-time performance, which was empowered by lightweight segmentation algorithms.

In [178], Perez-Jimenez et al. (2020) presented a UAV positioning system adroit at inspecting electric lines in GPS-denied environments, by means of enabling a range of inspection and maintenance tasks. Zhao et al. (2023), in [179], presented the optimization of UAV flight and communication energy consumption for autonomous power line patrols, highlighting the associated safety and efficiency benefits. The research by Cantieri et al. (2020) in [180] focused on using UAVs for the autonomous inspection of power distribution lines, addressing challenges like precise positioning, obstacle detection, and control.

4.2.5. AI-Driven UAV Systems for Power Line Inspections

Scholars have developed very advanced UAV systems equipped with advanced sensors and AI-driven data processing to ensure precise, efficient, and safe inspections of power lines. For instance, in [181], Moreno et al. (2022) introduced a Hybrid Fixed-Flapping Wing UAV for power line inspections, equipped with advanced avionics, cameras, and autonomous flight capabilities. Data acquisition included onboard avionics, a wattmeter, NanoPi, and motion capture systems for precise power consumption measurement and performance evaluation, utilizing data processing algorithms for accurate assessment. An AI-powered UAV system for urban power line inspection in Brazil was presented by Rangel et al. (2022) in [182], using drones with various sensors for data acquisition. Onboard AI processes images, recognizes patterns, and facilitates autonomous navigation through SIGI software, generating GIS-based maps and maintenance databases using data analysis algorithms.

Tsellou et al. (2023) and Du et al. (2022) highlighted UAVs’ role in enhancing fault detection and transmission line inspections using RGB and thermal sensors. Their data analysis involved image segmentation and machine learning algorithms to locate power lines and identify faults [183], with data transmitted to ground systems for navigation through the UAV equipped with various sensors and cameras for data collection [184]. UAVs for precise power line inspections with high-precision spatial data acquisition and real-time processing were emphasized in the studies by Liu et al. (2021) and Medrano et al. (2020). The data underwent denoising and segmentation algorithms for accuracy, and an Autonomous Data Acquisition System (ADAS) employed computer vision algorithms for high-quality visual data capture [185] [186]. UAV applications in improving efficiency and safety in power line inspections were discussed by Constantin and Dinculescu (2019) in [187]. Sensor-equipped UAVs gather detailed power lines and solar panel information, with data processing and analysis algorithms identifying maintenance needs and efficiently planning repairs.

4.3. Related Trend Analysis

To explore research trends, VOS-Viewer software (version 1.6.19.0) was used to analyze the 75 selected papers through a structured approach. This study applied co-occurrence analysis, examining how frequently keywords appeared together. A full counting method was chosen, making sure that all keyword occurrences were counted equally. The analysis considered all keywords from the papers, but to highlight the most relevant terms, a minimum threshold of three occurrences was set. Out of 749 unique keywords, 44 met this threshold. For these, the total strength of their co-occurrence links with other keywords was calculated. The keywords with the highest link strength were then selected for further analysis, offering a clear picture of key research themes.

4.3.1. Cluster Analysis of UAV Applications in Inspection Tasks

Table 5. Clusters of keywords, total link strength, and occurrences.

Description	Keywords	Total Link Strength	Occurrences
Cluster 1 (Red): 13 Keywords	aircraft detection	208	29
	convolutional neural networks	49	6
	deep learning	152	21
	defects	37	4
	fault detection	36	6
	image processing	63	7
	inspection methods	21	3
	learning systems	63	6
	line detection	34	4
	manual inspection	33	4
	neural networks	24	3
	power line detection	39	4
	signal detection	27	3
Cluster 2 (Green): 12 Keywords	air navigation	33	4
	cameras	31	5
	data handling	19	3
	electric power transmission networks	139	18
	image enhancement	25	4
	inspection	266	38
	inspection image	41	4
	power	51	7
	power line inspections	319	54
	power lines	252	39
	transmission line	84	11
	vehicle transmissions	50	6
Cluster 3 (Blue): 9 Keywords	antennas	296	44
	flight control systems	24	5
	image segmentation	25	4
	inspection and maintenance	17	3
	maintenance	27	4
	mapping	18	3
	robotics	34	7
	semantic segmentation	50	7
	unmanned aerial vehicles	396	67
Cluster 4 (Yellow): 6 Keywords	aerial vehicle	84	10
	deep reinforcement learning	17	3
	inspection process	22	3
	inspection tasks	25	3
	learning algorithms	36	5
	reinforcement learning	23	3
Cluster 5 (Purple): 4 Keywords	object recognition	40	5
	object detection	54	6
	outages	19	3
	power supply company	27	3

presents keywords grouped into five clusters, reflecting different research directions in UAV-based inspections based on their occurrences and total link strength. In Cluster 1 (red), the focus is on AI-driven fault detection, where “aircraft detection” appears 29 times, with a 208 link strength, and “deep learning” appears 21 times, with a 152 link strength. These numbers indicate that deep learning is a key tool in UAV-based fault identification and image analysis. Cluster 2 (green) highlights UAV applications in power infrastructure, with “power line inspections” having 54 occurrences and a 319 link strength, while “inspection” appears 38 times, with a 266 link strength. This suggests that UAVs are widely researched for automating power grid monitoring and improving maintenance strategies. Cluster 3 (blue) focuses on UAV technology and control, where “unmanned aerial vehicles” dominate, with 67 occurrences and a 396 link strength, followed by “antennas”, with 44 occurrences and a 296 link strength. The high numbers here suggest ongoing advancements in UAV flight control, navigation, and communication systems.

Cluster 4 (Yellow) examines UAV-based inspection methods, where “aerial vehicle” appears 10 times, with an 84 link strength, and “deep reinforcement learning” appears 3 times, with a 17 link strength. Although these numbers are lower, they show an emerging interest in making UAVs more autonomous. Cluster 5 (purple) relates to object detection and infrastructure monitoring, where “objects detection” has six occurrences, with a 54 link strength, and “power supply company” appears three times, with a 27 link strength. This shows that UAVs are being explored for asset tracking, outage detection, and security applications. The overall observation is an indication that the keyword distribution shows a clear trend toward integrating AI in UAV inspections, particularly in power transmission and defect detection. The strong presence of deep learning and power line inspections suggests these are central research areas, while the emerging focus on reinforcement learning indicates a shift toward more autonomous UAV operations. Future studies will likely enhance UAV decision-making, improve AI-driven analytics, and expand UAV applications for large-scale infrastructure maintenance.

4.3.2. Network Map Visualization Analysis of UAV Applications in Inspection Tasks

Figure 3 visually represents how research topics related to UAV-based power line inspections are connected. The larger keywords indicate topics that appear more frequently in the studies, with “unmanned aerial vehicles (UAVs)”, “inspection”, and “deep learning” being the most common. The thickness of the lines between the keywords shows how closely they are related, where the strong link between “deep learning”, “power line detection”, and “image processing” suggests that researchers are focusing on AI-driven fault detection. The different colors separate major research themes. The red group studies how AI can help UAVs inspect images. The key terms include “convolutional neural networks” and “neural networks”. The green group looks at how UAVs are used in power networks. The important terms are “power line inspections” and “electric power transmission networks”, which show how UAVs aid in monitoring and maintenance. The blue group focuses on improving UAV technology, featuring “antennas”, “flight control systems”, and “robotics”, which enhance navigation and performance.

The yellow group examines how to operate UAVs autonomously. This group uses “deep reinforcement learning” and studies the “inspection process” to improve AI-controlled drones. Collectively, the groups demonstrate a strong connection between UAVs and AI, paving the way for a more efficient future. Furthermore, the purple group highlights object detection, with “object detection” and “outages”, suggesting that UAVs are used to identify faults and improve grid reliability. This network of research topics suggests that UAVs are increasingly used in power line inspections, with AI being a key factor in improving automation and fault detection. The strong links between different areas indicate that research is moving toward making UAVs more autonomous, enhancing real-time monitoring, and refining data analysis techniques for better power grid management.

4.3.3. Overlay Map Visualization Analysis of UAV Applications in Inspection Tasks

Figure 4 provides an overlay map of research topics in UAV-based power line inspections from 2019 to 2023, where the color gradient indicates how the research focus has changed over time. Earlier studies, shown in blue and dark green, which were published between 2019 and 2020, concentrated on UAV technology fundamentals, with keywords like “flight control systems”, “robotics”, and “convolutional neural networks” reflecting the development of UAV hardware and AI-based image processing. In the green shades, from 2021 onward, there is a noticeable shift toward UAV applications in infrastructure monitoring, with increasing research on “transmission line”, “image enhancement”, and “power line detection”, indicating an increasing trend toward practical deployment. Recent research, shown in yellow, conducted between 2022 and 2023, indicates a growing focus on AI-driven automation, as seen in keywords like “deep reinforcement learning”, “inspection process”, and “learning algorithms”, indicating efforts to make UAV inspections more autonomous and predictive. The observations show that large nodes like “unmanned aerial vehicles (UAV)” and “deep learning” remain central across all years, confirming that AI integration continues to drive UAV advancements. The increasing presence of “inspection process” and “learning algorithms” between 2022 and 2023 suggests a trend toward real-time, AI-powered decision-making in UAV-based inspections. Future research is expected to enhance UAV autonomy, improve fault detection accuracy, and develop more efficient data-driven inspection methods for power grid monitoring.

5. Overall Discussion and Observations

This review highlights progress in power line inspections that are based on UAVs while pointing out several areas that still need improvement. AI techniques, such as CNNs, LiDAR, and federated learning, have improved fault detection, but challenges remain, particularly in real-time identification under extreme weather conditions. Image processing techniques like reinforcement learning and GAN-based deblurring have improved UAV capabilities, but large-scale multi-UAV coordination is still lacking. Another major challenge is electromagnetic interference, which affects UAV performance, along with battery life limitations that restrict flight duration. These issues suggest a need for EMI-resistant UAVs and better power solutions, such as wireless charging or energy-efficient navigation systems.

The cluster analysis of UAV applications shows that research is largely focused on AI-powered inspections, power infrastructure monitoring, UAV navigation, and object detection. Deep learning has become a dominant tool for power line inspections, but newer research is exploring reinforcement learning to improve UAV autonomy. One of the main gaps in this area is the integration of UAVs into predictive maintenance and smart grid systems. Enhancing UAV performance in extreme weather conditions and ensuring communication that is reliable between UAVs and power networks are ongoing concerns and need more attention.

The network analysis reveals strong links between deep learning, power line detection, and image processing, highlighting AI’s growing role in UAV inspections. However, a complete integration of predictive maintenance systems and asset management is still a lacking area of consideration. In making UAV inspections more effective, improvements are needed in UAV communication networks, AI-driven mission planning, and real-time monitoring capabilities. While UAVs are being used for outage detection and infrastructure security, their ability to perform large-scale inspections autonomously remains an area for further development.

The overlay map analysis shows a shift in research priorities over time. Earlier research from 2019 to 2020 focused mostly on AI-based image processing and on UAV hardware; however, recent research from 2022 to 2023 has moved toward predictive analytics and automation. The increased emphasis on real-time learning algorithms specifies a trend toward proactive fault detection. Moving forward, research should pay attention to improving onboard AI processing, evolving coordinated UAV fleets for large-scale inspections, and refining UAV resilience in harsh environments. Addressing these challenges will improve UAV effectiveness in power grid maintenance and long-term asset monitoring.

6. Conclusions and Proposed Future Research Directions

This paper systematically reviewed the application of UAVs for power line inspections using the PRISMA methodology to ensure a structured and comprehensive analysis. The Scopus database was utilized to identify relevant publications, resulting in the selection of 75 documents after screening and eligibility assessment. Among other things, this study included bibliometric analysis, covering the top 10 most cited journals and conference proceedings and the top 10 most cited publications in the field. Additionally, this review examined innovative advancements such as UAV-based fault detection strategies and algorithms, UAV-based monitoring challenges, robotic vision, UAV autonomous technologies, and AI-driven UAV systems. Thereafter, emerging trends and the gaps in the existing literature were discussed. Notably, advancements in predictive maintenance, multi-UAV operations, and real-time data analysis were observed. However, significant challenges remain, including operational issues, resilience to extreme weather, and large-scale automation.

Finally, future research directions to improve UAV-based power line inspections are proposed as follows:

• While image processing techniques like reinforcement learning and GAN-based deblurring have enhanced UAV capabilities, effective coordination of multiple UAVs for large-scale inspections is still lacking. Research should explore advanced swarm intelligence and cooperative control algorithms.
• UAV performance is affected by EMI, which can disrupt navigation and data transmission. There is a need for EMI-resistant UAVs and improved shielding techniques to enhance operational reliability.
• Limited battery life affects UAV flight duration, posing a significant challenge for long-range power line inspections. Research should explore wireless charging solutions, energy-efficient navigation, and hybrid power sources to extend UAV endurance.
• UAVs have yet to be fully integrated into predictive maintenance frameworks and smart grid operations. Future studies should focus on seamless data exchange, interoperability with grid monitoring systems, and AI-driven decision-making.
• The shift toward predictive analytics and automation highlights the importance of enhancing onboard AI processing capabilities. Future research should focus on real-time edge computing and adaptive learning algorithms to improve UAV autonomy in inspections.
• UAVs must be adapted to function effectively in extreme conditions, such as high winds, heavy rainfall, and fluctuating temperatures. Research should explore materials, aerodynamics, and sensor technologies that improve UAV durability and operational reliability.