Unmanned Aerial Vehicles in Last-Mile Parcel Delivery: A State-of-the-Art Review

Abstract

Unmanned Aerial Vehicles (UAVs) are being increasingly implemented in parcel delivery applications. The scientific progress in this field is progressing exponentially. However, there is a notable gap in synthesizing recent research progress in UAV applications for last-mile delivery. This review study addresses this gap and conducts an in-depth review of UAV research for last-mile delivery across seven domains: environmental performance, economic impacts, social impacts, policy and regulations, routing and scheduling, charging infrastructure, and energy consumption. The review indicates that UAVs promise to reduce last-mile delivery emissions by 71% and costs by 96.5% compared to truck delivery. Saturated knowledge analysis is conducted across the seven domains to identify potential research gaps. Additionally, this review identifies key knowledge gaps, including variability in environmental and cost data, limitations associated with 2D modelling, and a lack of experimental validation. Future research interventions aimed at advancing UAV adoption in last-mile delivery applications are discussed.

1. Introduction and Background

Electronic commerce (e-commerce) has revolutionized the shopping experience, offering a more convenient and accessible process [1]. It represented 18.8% of the total retail sales worldwide in 2022 and is estimated to rise to 22.6% by 2027 [2]. The COVID-19 pandemic accelerated e-commerce activities, causing a 150% surge in parcel volumes [3]. As such, parcel delivery volumes increased from 64 billion in 2016 to 161 billion in 2022 and are projected to reach 225 billion by 2028 [4].

The rapid growth in e-commerce activities is coupled with higher consumer expectations for expedited delivery and lower shipping costs [5,6,7,8]. It is reported that 80% of online customers in the US prefer same-day delivery [9]. In this respect, truck delivery systems are falling short of satisfying these expectations, as shown by the increasing volume of unfulfilled deliveries, ranging from 10% to 15% in the US alone [10]. Furthermore, truck delivery induces additional costs for expedited delivery options (e.g., same-day delivery).

Recently, it has been indicated that 86% of last-mile deliveries are made for parcels weighing less than 5 pounds [11,12]. Therefore, Unmanned Aerial Vehicles (UAVs) are emerging as a viable alternative to truck delivery [7,13,14]. UAVs provide faster, more flexible service with reduced costs and lower environmental impacts [3,5,6,7]. Furthermore, UAVs diminish the need for trucks, reducing congestion and the associated health, economic, and environmental externalities [6,15]. However, some UAV applications in last-mile delivery operate in coordination with trucks, which is known as “truck-assisted UAV delivery” or the “flying sidekick model” in which UAVs are launched from moving vehicles to serve nearby delivery points and then return to the vehicle [16,17].

Research on using UAVs as a viable last-mile delivery solution is progressing hastily [18,19]. UAVs are expected to be responsible for 40% of deliveries by 2030 [7,20,21]. However, multiple limitations, constraints, and public concerns need to be addressed for the broad adoption of UAVs.

Previous studies provide diverse assessments of UAV capabilities for last-mile delivery applications across multiple research domains [6,7,13,15,22]. Indeed, the literature on UAV last-mile delivery is rich, and substantial contributions have been produced. However, UAV applications in last-mile delivery entail the seamless integration of various research domains [22,23]. As such, there is a need for a review study that synthesizes various research efforts on UAV applications in last-mile delivery operations. Surprisingly, previous review attempts (see

Table 1. Overview of previous UAV review studies.

Studies	Environmental Performance	Economic Impacts	Social Impacts	Policy and Regulations	Routing and Scheduling	Charging Infrastructure	Energy Consumption
Stöcker [24]				✓
Coutinho [25]					✓
Thibbotuwawa [26]					✓
Macrina [27]					✓
Benarbia and Kyamakya [6]					✓
Moshref-Javadi and Winkenbach [28]					✓
Smith [29]			✓
Pasha [30]					✓
Garg [7]	✓	✓
Li [13]	✓		✓		✓
Eskandaripour and Boldsaikhan [22]	✓				✓		✓
Raivi [5]					✓
Jazairy [31]	✓	✓	✓
Present study	✓	✓	✓	✓	✓	✓	✓

), albeit offering significant contributions, focused on a few domains, signifying the need for a comprehensive review of UAV last-mile delivery applications, as depicted in Table 1.

1.1. Scope Identification

The scope of this review study is defined using three steps: relevancy, sampling, and domain identification. First, 98 relevant articles published during the past 10 years were identified. The sample is sourced from three primary databases, including Science Direct, Scopus, and Web of Science, using several keywords.

Following an initial scoping analysis, seven domains are identified as the primary focus of these studies. Furthermore, 19 secondary topics also emerged, forming the base of this detailed review. The largest proportion of studies (39.4%) focus on routing and scheduling, followed by economic impacts (21%) and environmental impacts (19.2%). Policy and regulations are the least explored domain (8.2%), as depicted in Figure 1.

It should be noted that the present study contrasts the features of UAV last-mile delivery to those of truck-assisted UAVs and truck-based deliveries. Other last-mile delivery modes, such as human, bike, or cart-based last-mile delivery solutions, are not discussed within the present study.

1.2. Detailed Contribution

The present study synthesizes the current literature on UAV last-mile delivery. In particular, the study focuses on seven research domains, encompassing 19 research topics relevant to the implementation of UAVs in last-mile delivery, as detailed in Figure 2. These domains have never been jointly reviewed in previous research.

In particular, the contributions of the present study include

• Developing a holistic understanding and assessment of utilizing UAVs in last-mile delivery;
• Identifying saturated knowledge domains in UAVs' last-mile delivery research;
• Identifying knowledge gaps in UAVs' last-mile delivery research;
• Suggesting avenues for future research interventions informing academia, practice, and policy.

2. UAV Last-Mile Delivery: A Detailed Review

2.1. Environmental Performance

The environmental benefits of UAVs are promoted as the main motivation for their last-mile delivery applications [31,32]. Environmental assessment is introduced in the literature for utilization and production/disposal phases [32,33]. The utilization phase includes the generated emissions from two stages: generation-to-battery (GTB) and battery-to-propeller (BTP). The production/disposal phase includes emissions from components and battery manufacturing and recycling [34].

Acknowledging the varied operational contexts and their impacts on the environmental performance of UAVs, the average GHG emissions synthesized from the literature include the utilization phase (representing ≈ 87.5% to 97.5%) and battery production/disposal (≈12.5% to 2.5) [34,35]. However, Koiwanit [36] reported that the production/disposal phase represents 99.2% of the total delivery emissions. This discrepancy may be attributed to Koiwanit’s consideration of producing all UAV components rather than focusing solely on the battery.

Typically, GHG emissions are reported per vehicle (UAV) kilometre travelled (VKT) and range between 1.9 and 79 g Co2eq/km. In comparison, GHG emissions per parcel average between 0.6 and 1786 g Co2eq per parcel [34,36,37,38]. This wide range is attributed to several parameters. For example, regulation strictness increases emissions by up to 400% [38], while a 50% reduction in demand density (e.g., 500 to 250 recipients/mile) reduces GHG emissions by up to 84% [39]. Similarly, a higher electricity carbon intensity (399 to 900 g Co2eq/kWh) increases GHG emissions by up to 93% [35,40].

Compared to diesel and electric trucks (e-trucks), UAVs reduce emissions by up to 71% and 69%, respectively [31], as detailed in

Table 2. UAV environmental performance assessment in the literature.

Studies	GHG Emissions				Operation Parameters
	UAVs	Diesel Trucks	Electric Trucks	Unit (g Co2eq per)	Electricity Carbon Intensity (g Co2eq/kWh)	Demand Density	Payload (kg)
Koiwanit [36]	79.0	NA	NA	Km	NA	Rural	5.0
Goodchild and Toy [39]	13.0	740.0	NA	Km	377.3	NA	NA
Rodrigues [35]	12.0	931.5	782.5	Km	399.0	NA	0.5
	23.5				900.0
	48.5	NA	NA	Delivery	399.0
	93.8				900.0
Park [37]	1.9	NA	NA	Km	Low	NA	NA
	4.4				High
	3.5			Delivery		Urban
	39.0					Rural
Figliozzi [42]	160.0	30.0	NA	Delivery	560.0	NA	5.0
Stolaroff [34]	290.0	1015.0	943.0	Delivery	200.0	NA	0.5
	915.0				1000.0
	578.0				200.0		8.0
	1786.0				1000.0
Raghunatha [40] *	6.0	5.0	1.0	Delivery	47.0	Urban	544.3
	16.0	22.0	4.0			Rural
Elsayed and Mohamed [38]	0.6	100.0	2.0	Delivery	52.6	Urban	2.3
	1.5
	24.0	1900.0	17.0			Rural
	25.0
Baldisseri [33] **	NA	730.0	675.0	Delivery	NA	NA	2.5
Chiang [16] **	NA	1811.0	NA	Delivery	NA	Urban	2.3
		2101.0				Rural
Li [41] **	NA	1718.0	NA	Delivery	NA	Urban	2.3
		2339.0				Rural

* UAV type is eVTOL. ** Considered truck-assisted UAV scenario.

. Furthermore, truck-assisted UAVs reduce the number of trucks by up to 33%, resulting in 28.6% emission reductions compared to diesel trucks. This integration also reduces GHG emissions by up to 20% compared to e-trucks. Notably, the range of truck-assisted UAVs is between 539 and 2142 g Co2eq per delivery [16,33,41]. Table 2 demonstrates a significant variation in the GHGs emitted by UAVs in last-mile delivery. The lowest emissions (0.6 g Co2eq/delivery) result from a UAV delivery in an urban area with lean flight regulations (Lean flight regulations permit greater operational freedom by allowing higher flight altitudes, reduced buffer zones around people and structures, and more flexible airspace usage.) [11,38], low carbon intensity (52.6 g Co2eq/kWh), and a payload capacity of 2.25 kg/UAV [38]. In contrast, the highest emissions (1786 g Co2eq/delivery) result from a UAV delivery in an urban area with high carbon intensity (1000 g Co2eq/kWh) and a payload capacity of 8 kg/UAV [34]. However, 87.5% to 97.5% of these emissions could be eliminated using renewable energy resources to power UAVs in last-mile delivery [37].

The wide variation in GHG emissions is attributed to the unique characteristics of each study, whereby external factors (e.g., payload, wind, speed, etc.) have a significant bearing on GHG emissions. However, based on the observed variation, there is a dire need for an accurate estimation of GHG emissions for the utilization and production/disposal phases of UAV delivery. In addition, a sensitivity analysis that quantifies the contributing factors to GHG emissions is needed. Comparisons between different delivery methods (e.g., UAVs, truck-assisted UAVs, and e-trucks) would be beneficial in comprehensively evaluating the environmental benefits of UAVs across various contexts.

2.2. Economic Impacts

Economic assessment in last-mile delivery is carried out in two phases: economic viability/return and delivery cost. It is reported that UAVs enhance economic return by up to 229% depending on the operational context [15,43], while the average delivery cost synthesized from the literature includes the fleet cost (≈32%), charging station cost (≈8%), and operational cost (≈60%) [11] and is reported to be 96.5% lower than that of truck delivery [44].

In non-business-to-consumer contexts, such as healthcare, UAVs have been found to significantly reduce transit times (e.g., 70% in health services logistics). However, they remain more expensive than trucks unless operation costs are cut by over 80% [45]. This highlights the difficulty of achieving economic viability in small-scale, urgent logistics scenarios.

Overall, the cost per parcel delivered, often used as the cost unit, ranges between USD 0.03 and USD 0.67 (Table 3). This wide range is attributed to several parameters. Extending UAV operating hours is reported to reduce the cost to USD 0.2 [17,40,46]. Similarly, the cost per parcel in higher demand density (e.g., urban contexts) is reported as USD 0.07 compared to USD 0.67 in rural contexts [44], as detailed in Table 3. Compared to truck delivery, the cost reduction caused by UAVs in last-mile delivery is attributed to reducing human resource costs (44–70% of the total cost) compared to 6% for UAV last-mile delivery [40,44].

In truck-assisted UAV delivery, the cost per parcel ranges from USD 0.48 to USD 6.5 [17,41]. It is argued that truck-assisted UAV delivery is the most economically viable option [17]. They also argue that UAVs must charge at the depot, significantly increasing the number of required UAVs to serve the demand, which increases the cost per parcel compared to a truck-assisted UAV system. However, this statement is not echoed in the literature.

Table 3. UAVs’ economic performance.

Studies	Cost per Parcel (USD)			Operation Parameters
	UAVs	Diesel Trucks	Electric Trucks	Regulations	Operating Hours (hr)	Demand Density	Payload (kg)
Sudbury and Hutchinson [46]	0.40	1.20	NA	NA	12	Urban	2.3
Elsayed and Mohamed [44]	0.03	2.00	1.30	Lean regulations	12	Urban	2.3
	0.07			Strict regulations
	0.47	15.00	NA	Lean regulations		Rural
	0.67			Strict regulations
Choi and Schonfeld [17] *	0.55	0.57	NA	NA	12	Urban	5.5
ElSayed [11] **	0.08	NA	NA	Lean regulations	6	Urban and Rural	2.0
	0.09			Strict regulations
Chiang [16] *	NA	5.40	NA	NA	8	Urban	2.3
		6.70				Rural
Li [41] *	NA	6.50	NA	NA	8	Urban	2.3
		6.90				Rural
Raghunatha [40] ***	0.03	0.07	0.07	NA	8	Urban	544.3
	0.07	0.18	0.16			Rural
Borghetti [43] **	0.25	1.10	NA	NA	13	Urban	2.3
Haidari [47]	0.33	0.41	NA	NA	NA	Urban	3.0

* Considered a truck-assisted UAV scenario. ** Considered the cost of charging stations. *** The UAV type is eVTOL.

Table 3. UAVs’ economic performance.

Studies	Cost per Parcel (USD)			Operation Parameters
	UAVs	Diesel Trucks	Electric Trucks	Regulations	Operating Hours (hr)	Demand Density	Payload (kg)
Sudbury and Hutchinson [46]	0.40	1.20	NA	NA	12	Urban	2.3
Elsayed and Mohamed [44]	0.03	2.00	1.30	Lean regulations	12	Urban	2.3
	0.07			Strict regulations
	0.47	15.00	NA	Lean regulations		Rural
	0.67			Strict regulations
Choi and Schonfeld [17] *	0.55	0.57	NA	NA	12	Urban	5.5
ElSayed [11] **	0.08	NA	NA	Lean regulations	6	Urban and Rural	2.0
	0.09			Strict regulations
Chiang [16] *	NA	5.40	NA	NA	8	Urban	2.3
		6.70				Rural
Li [41] *	NA	6.50	NA	NA	8	Urban	2.3
		6.90				Rural
Raghunatha [40] ***	0.03	0.07	0.07	NA	8	Urban	544.3
	0.07	0.18	0.16			Rural
Borghetti [43] **	0.25	1.10	NA	NA	13	Urban	2.3
Haidari [47]	0.33	0.41	NA	NA	NA	Urban	3.0

* Considered a truck-assisted UAV scenario. ** Considered the cost of charging stations. *** The UAV type is eVTOL.

From the literature, we synthesized the optimal economic utilization for UAV last-mile delivery. Figure 3 indicates that the lowest cost (0.03 USD/parcel) occurs in urban areas with lean flight regulations, 12 operating hours, and a payload capacity of 2.25 kg/UAV. In contrast, the highest cost (0.67 USD/parcel) occurs in rural areas with strict flight regulations.

Reports from real-world UAV operations provide operational details that are not often captured in the academic literature. For example, the Zipline home delivers supplies within up to a 16 km delivery radius, typically with parcels weighing 3.5 kg [48]. Similarly, Manna Aero reports delivering orders with UAVs with customer fees of around EUR 1.99 per trip, noting that each flight is currently profitable over a 3.4 km range [49]. Wing recently introduced a UAV model capable of delivering parcels weighing up to 2.3 kg across a 19.3 km round trip [50]. These services illustrate key operational parameters, such as payload capacity, range, and pricing, which are critical for evaluating UAV-based last-mile delivery systems.

Despite the growing interest and promising economic potential, significant variations in cost estimates limit the applicability of UAVs in last-mile delivery. Consequently, a comprehensive cost assessment is needed to clarify the significance of different cost elements. Additionally, sensitivity analyses quantifying the impact of cost variables across various delivery methods would provide valuable insights to evaluate the economic benefits of UAVs in different contexts. Considering indirect savings, such as reduced congestion from replacing truck deliveries, would enhance UAV system evaluations.

2.3. Social Impacts

The social impacts of UAV applications in last-mile delivery are yet to be fully defined. However, it is approximated using three integrated measures: (1) public acceptance, defined as the attitude towards the use of UAVs [51]; (2) safety, which relates to the risk of accidents or harm resulting from UAV operation [13]; and (3) privacy, defined as the lack of disruption and/or monitoring by others [52].

The literature highlights public acceptance as essential in integrating UAVs [29]. However, there is a significant variation in public acceptance of UAV applications across countries [32,53,54,55], socio-economic and demographic (SED) profiles [53,54], and geography (urban vs. rural) [53,54]. Moreover, there is evidence of contradictory expectations from business owners and end-users. While UAVs facilitate flexible delivery times and accessibility to distant locations for businesses, end-users are worried about additional delivery fees [32]. Hence, they have a lower acceptance of using UAVs in delivery from an economic perspective.

The variation in public acceptance is interlinked with several factors; the most notable is UAV regulations. The permission to overflight is reported to enhance public acceptance and vice versa. In Germany, UAV overflight is not permitted, and public acceptance of UAVs is relatively low [53]. However, in Canada and the USA, regulations guide overflight, and there is relatively higher public acceptance [56]. Combined, these factors significantly impact the degree of public acceptance. For example, the USA [54], Singapore [55], and Portugal [32] reported higher public acceptance, yet significant variation was observed between rural and urban contexts [53,54]. The latter is more open to UAV delivery. In contrast, results from Switzerland [55] and Germany [53] indicate lower public acceptance and highlight variations in different SED groups.

Furthermore, the literature indicates that UAV applications in last-mile delivery are perceived unsafe by the public. There are concerns about the misuse of UAVs in illegal applications, such as drug and weapon trafficking, and expected incidents (e.g., payload dropping) [3,53,54,55]. Safety perception varies with public knowledge of UAV operations [53], their regulatory environment, and prior experience/exposure. Reacting to that, international institutions (e.g., the Federal Aviation Association (FAA) and European Commission) are actively developing safety-oriented bylaws and regulations [53,54] to enable applications of UAVs in delivery.

The literature shows that the public considers UAV applications in last-mile delivery to be a breach of privacy. Major concerns include the ability to collect, retain, use, and disclose personal information [3,52]. For instance, community opposition in Canberra (Australia) led to the termination of Google Wing’s UAV delivery trial, highlighting that technological solutions must align with social expectations [57].

Public concerns vary across various aspects, including public knowledge, experience level, UAV speed, and data collection capabilities [29,54,58]. Public concerns can vary significantly based on the extent of UAV adoption.

Table 4. Measures of UAVs’ social impacts.

Measures of Social Impact	Range Reported in the Literature (%)	Measurement Method	Notable Research Gaps
Public acceptance	(18–83%)	Survey	- Lack of experimental research - Lack of direct contact with UAVs during operation - Reliance on revealed preferences surveys
Safety concerns	(18–91%)	Survey
Privacy concerns	(30–86%)	Survey

summarizes findings and gaps associated with the social impacts of UAVs in last-mile delivery.

Previous research indicates that the public is ill-informed about using UAVs in delivery [29]. In addition, research on assessing the social impacts of UAVs in last-mile delivery depends primarily on stated preferences data collected from surveys (examples of survey measures are detailed in Appendix A.1). However, using different methodological approaches in UAV surveys causes discrepancies in the results [29]. Also, survey participants did not observe UAV operation in action nor the associated noise, size, trajectory, and even colour.

Furthermore, the public has yet to consider the safety and privacy concerns of using UAVs on a large scale, given the limitations in current UAV operations [59]. As such, the literature emphasizes the dire need to assess/validate social impacts through experimental research. We echo this argument and call for experimental research to identify the thresholds of UAVs’ social impacts.

2.4. Policy and Regulations

UAVs are relatively new, and concerns about possible risks to individuals, properties, and other airspace users necessitate the development of a comprehensive regulatory framework [38]. Several countries are formulating UAV regulations, aiming to (1) regulate the use of airspace, ensuring the safe coexistence of UAVs and manned aircraft; (2) establish operational boundaries, guaranteeing responsible and safe operation of UAVs; And (3) establish administrative procedures, including flight permissions, pilot licencing, and data collection authorization [24].

UAV last-mile delivery applications are regulated across four primary dimensions. Operational limitations involve constraints applied to UAV flights outlining essential flight parameters. Regulatory classifications involve a set of categories (e.g., weight and size) for UAVs. Administrative frameworks include policies regulating flight permissions and procedures. Ethical consideration involves rules managing privacy and data protection [24,60,61]. Previous studies highlight five operational limitations.

Table 5. Operational limitations (sources: Appendix A.2).

Criteria	Regulation Category
	Lean or Unrestricted (Countries)	Average (Countries)	Strict (Countries)
Distance from airports or strips	No-fly zone (Netherlands, Austria, Germany, Japan, Malaysia, UK, China, and France)	2–5 km (Austria, Italy, Chile, Colombia, and Canada)	8–10 km (US, Spain, South Africa, and Rwanda)
Distance from property and people	Restricted over crowds (US, Spain, Malaysia, France, Germany, China, Austria, Italy, and the Netherlands)	30 m (Canada, Australia, Chile, and Japan)	50 m (UK, South Africa, Colombia, and Rwanda)
Distance from congested area/highway	<150 m (Canada and Germany)	150 m (UK)	Not allowed (Australia)
Maximum altitude	130–155 m (France, Nigeria, Austria, Italy, Colombia, Japan, and Chile)	100–122 m (UK, Spain, Australia, Malaysia, US, Canada, the Netherlands, and China)	≤100 m (Germany and Rwanda)
VLOS	>500 m (Australia, Malaysia, Germany, Austria, Azerbaijan, Japan, Nigeria, China, and Colombia)	500 m (or) EVLOS (UK, US, the Netherlands, Italy, Spain, and Chile)	≤300 m (France and Rwanda)
BVLOS	Allowed (France, Nigeria, and Rwanda)	Special approval needed (UK, Australia, Malaysia, the Netherlands, Germany, Italy, Austria, Spain, Azerbaijan, South Africa, Japan, and China)	Not allowed (Chile and Colombia)

presents three regulatory classifications and the associated parameters across various operational indicators [11,24,38,62,63].

The literature highlights that regulations substantially impact UAVs’ last-mile operations, especially in urban contexts. Choosing strict over lean regulations enhances safety but reduces routing efficiency, leading to increased energy consumption and 75% more failed delivery missions as sections of the urban volume become inaccessible. Addressing mission failures necessitates a 223% increase in charging stations compared to lean flight regulations.

In addition, strict flight regulations increase costs and GHG emissions by up to 10% and 400%, respectively [11,38]. Nevertheless, implementing lean flight regulations requires advanced technical prerequisites, including geofencing and collision avoidance features, to achieve the necessary level of safety [60]. Additionally, many medical delivery items (e.g., vaccines, blood, and biological samples) are classified as dangerous goods (DGs) under air transport regulations. Grote et al. (2021) audit these dangerous goods rules and emphasize that they were developed for crewed aircraft and do not adequately cover UAV operations [64]. They conclude that new governance frameworks will be required to allow for safe UAV delivery of DG-classified medical cargo.

Similarly, ElSayed [11] cross-examined regulations for both horizontal proximity to buildings (H) and permissible flight altitude (A), with each categorized as lean (1), average (2), or strict (3), resulting in nine scenarios (e.g., H1A1 for lean horizontal proximity and flight altitude). The findings did not echo the same wide range of variation for GHG emissions’ sensitivity to regulations; however, more than a 10% increase in cost is reported, as illustrated in Figure 4.

Several aspects of policy and regulation have yet to be discussed in the literature. Most notable is the lack of regulatory frameworks for autonomous UAV last-mile delivery [11]. Furthermore, studies have identified noise as a critical constraint, yet there are no guidelines on acceptable UAV noise levels [60,66]. In addition, there is almost no mention of the interaction between UAVs and other road users, especially at the delivery/pick-up points, where UAVs will be flying at lower altitudes.

2.5. Routing and Scheduling

The routing and scheduling of UAVs are interconnected, and their definitions often overlap. Routing focuses on identifying optimal paths that meet UAV performance indicators within given constraints [67]. Scheduling integrates time-related factors, such as time windows and arrival times, into the routing problem [30,68]. Scheduling is also viewed as a task assignment problem, where specific delivery tasks are assigned to UAVs, aiming to minimize the number of UAVs needed to complete deliveries, optimizing resource utilization [25].

Routing and scheduling problems are based on the principles of the Travelling Salesman Problem (TSP) and Vehicle Routing Problem (VRP) [5,68,69,70]. TSP defines the shortest route that visits a set of locations once and returns to the origin point. At the same time, VRP extends this by considering multiple vehicles and additional constraints, such as delivery time windows and vehicle capacities [71]. Variants of these problems are frequently used in UAV routing and scheduling, including the following:

• Multi-Trip VRP (MTVRP): This reuses UAVs to overcome payload limitations [71,72].
• Capacitated VRP (CVRP): This considers the UAV payload capacity [16,25,26,71,73,74].
• Multi-Depot VRP (MDVRP): This involves multiple starting and ending points for UAVs [26,70].
• Time Window VRP (VRPTW): Each delivery location must be visited in a time window [26,70].
• Flying Sidekick TSP (FSTSP): UAVs and trucks deliver concurrently, with UAVs departing and returning to a truck following a TSP route [5,70,71,72,73,74,75,76,77].
• Parallel Drone Scheduling TSP (PDSTSP): UAVs and trucks simultaneously deliver along different paths from a common depot [70,72,74,75,78,79].

Optimal routing/scheduling also depends on the delivery method, number of UAVs, and parcel delivery volume [16,72,73,76,77,78,80,81]. Delivery methods include UAVs and assisted UAV systems, where UAVs work with trucks or public transport [69,82].

Furthermore, factors affecting routing and scheduling problems include the model environment (i.e., 2D and 3D) and UAV parameters (i.e., battery size, parcel weight, and flight altitude). Variants of 2D or 3D modelling coupled with static or dynamic obstacles significantly impact optimal routing [12,25,83,84,85]. Modelling in a 2D environment represents 77% of the surveyed literature, where the altitude is fixed or not specified [86,87]. Static environments are used in 61% of the surveyed literature, with the assumption of predefined delivery points and a focus on optimizing the sequence of deliveries [16]. Weather conditions are considered in 13% of the literature, while the impact of demand density on routing and scheduling is considered in 6% of the literature [25,26,80,88].

The performance of routing solutions is evaluated through metrics such as delivery cost, time, route distance, and emissions for path-planning problems [67]. In task assignment problems, performance is measured by the number of UAVs and trucks required to complete the delivery [25]. The objective of the two optimization problems is to maximize gains (i.e., UAV flight range) or minimize losses (i.e., parcel delivery time and costs).

Overall, routing and scheduling optimization problems are solved using several algorithms. Particularly, 65% of reviewed studies employ Heuristic and Genetic algorithms (e.g., A*, Dijkstra, Ant Colony Optimization (ACO), Genetic Algorithm, Simulated Annealing, and GRASP Metaheuristic). Table 6 shows the modelling features (e.g., number of UAVs, trucks, and parcels), modelling environments, and optimization approaches.

Despite advances in UAV routing and scheduling optimization, several research gaps hinder practical implementation. The focus on 2D environments (77%) overlooks the complexities of 3D UAV flights, which are essential for realistic path planning [89]. The interaction between multiple UAVs is not fully addressed, and experimental testing is mainly absent, limiting the validation of theoretical models. Also, only 13% of the surveyed studies consider weather conditions without considering the stochastic nature of wind and temperature.

Table 6. The literature on UAVs’ last-mile delivery routing and scheduling problems.

Studies	UAVs	Trucks	Parcels	Modelling Environment			Solution Algorithm
				2D	3D	Obstacles
Dorling [71]	Multiple		Multiple		✓		Simulated Annealing
Xu [90]	Multiple				✓	✓	Gradient Descent
Rovira-Sugranes and Razi [91]	Multiple			✓			Dijkstra
Agatz [69]	Single	Single	Single	✓			CPLEX
Chang and Lee [70]	Multiple	Single	Single	✓		✓	Clustering
Chen [82]	Single		Single	✓			A*
Torabbeigi [68]	Multiple		Multiple	✓			CPLEX
Kitjacharoenchai and Lee [73]	Multiple	Multiple	Multiple	✓			CPLEX
Chiang [16]	Multiple	Multiple	Single	✓			Genetic Algorithm
Dhulkefl and Durdu [86]	Single			✓		✓	A* and Dijkstra
Thibbotuwawa [88]	Multiple		Single		✓	✓	Clustering
Kuru [83]	Multiple		Multiple		✓		Hungarian and Cross-Entropy Monte Carlo
Wang [74]	Multiple	Multiple	Single	✓			Clustering
Cheng [72]	Multiple		Multiple	✓			Branch and Cut
Kim [80]	Single		Single	✓			Heuristic-Based
Huang [92] *	Single		Single	✓			CPLEX
Karve and Kapadia [87]	Multiple			✓			Dijkstra
Hazama [75]	Multiple	Single	Single	✓			Genetic
Huang [93] *	Single		Single	✓		✓	Dijkstra
Ghelichi [81]	Multiple		Single	✓			GUROBI
Sajid [76]	Multiple		Multiple	✓			Hybrid Genetic and Simulated Annealing (HGSA)
Gómez-Lagos [78]	Multiple		Single	✓			GRASP Metaheuristic
Li [85]	Single				✓	✓	A* Plus
Escribano [84]	Multiple		Multiple	✓		✓	CPLEX
Wang and Bai [67]	Single			✓			A*
Li [89]	Single				✓	✓	RJA*
Kumar [94]	Single			✓		✓	Rapid-A*
Zhao [77]	Multiple	Single	Multiple	✓			Genetic Algorithm
Zhu [79]	Multiple		Single	✓			Branch and Cut

* Considered a public transportation scenario.

Table 6. The literature on UAVs’ last-mile delivery routing and scheduling problems.

Studies	UAVs	Trucks	Parcels	Modelling Environment			Solution Algorithm
				2D	3D	Obstacles
Dorling [71]	Multiple		Multiple		✓		Simulated Annealing
Xu [90]	Multiple				✓	✓	Gradient Descent
Rovira-Sugranes and Razi [91]	Multiple			✓			Dijkstra
Agatz [69]	Single	Single	Single	✓			CPLEX
Chang and Lee [70]	Multiple	Single	Single	✓		✓	Clustering
Chen [82]	Single		Single	✓			A*
Torabbeigi [68]	Multiple		Multiple	✓			CPLEX
Kitjacharoenchai and Lee [73]	Multiple	Multiple	Multiple	✓			CPLEX
Chiang [16]	Multiple	Multiple	Single	✓			Genetic Algorithm
Dhulkefl and Durdu [86]	Single			✓		✓	A* and Dijkstra
Thibbotuwawa [88]	Multiple		Single		✓	✓	Clustering
Kuru [83]	Multiple		Multiple		✓		Hungarian and Cross-Entropy Monte Carlo
Wang [74]	Multiple	Multiple	Single	✓			Clustering
Cheng [72]	Multiple		Multiple	✓			Branch and Cut
Kim [80]	Single		Single	✓			Heuristic-Based
Huang [92] *	Single		Single	✓			CPLEX
Karve and Kapadia [87]	Multiple			✓			Dijkstra
Hazama [75]	Multiple	Single	Single	✓			Genetic
Huang [93] *	Single		Single	✓		✓	Dijkstra
Ghelichi [81]	Multiple		Single	✓			GUROBI
Sajid [76]	Multiple		Multiple	✓			Hybrid Genetic and Simulated Annealing (HGSA)
Gómez-Lagos [78]	Multiple		Single	✓			GRASP Metaheuristic
Li [85]	Single				✓	✓	A* Plus
Escribano [84]	Multiple		Multiple	✓		✓	CPLEX
Wang and Bai [67]	Single			✓			A*
Li [89]	Single				✓	✓	RJA*
Kumar [94]	Single			✓		✓	Rapid-A*
Zhao [77]	Multiple	Single	Multiple	✓			Genetic Algorithm
Zhu [79]	Multiple		Single	✓			Branch and Cut

* Considered a public transportation scenario.

2.6. Charging Infrastructure

The use of UAVs for last-mile delivery is limited by their short delivery range [95,96,97,98,99,100]. This limitation is countered by deploying charging stations (both stationary and mobile), which extend the operational range and efficiency of UAVs in last-mile delivery [99]. Stationary stations include those on rooftops or streetlights in urban areas or overhead power lines in rural areas [101,102]. Mobile stations utilize trucks or public transportation [100]. Charging methods include battery swapping and wireless or laser-beam recharging [102,103,104].

Some UAV warehouses incorporate charging capabilities as part of a broader infrastructure, including dispatch, weather shielding, and maintenance. This section focuses on charging infrastructure specifically, whether integrated within warehouses or implemented as standalone units.

The deployment of charging infrastructure is addressed through sophisticated optimization models that aim to determine optimal charging locations while maximizing coverage and minimizing the total cost [95,97,100]. Key factors influencing the UAV system configuration include the UAV flight range, regulation strictness, and model environment, as detailed in

Table 7. The literature on charging infrastructure.

Studies	Objective Function	Energy Consumption	Regulations	Modelling Environment	Solution Algorithm
Shavarani [95]	Minimize cost	Fixed	Not considered	2D	Genetic Algorithm
Hong [96]	Maximize coverage	Different with and without payload	Not considered	2D	Greedy Algorithm
Chauhan [97]	Maximize coverage	Function of payload and distance	Not considered	2D	Greedy Algorithm
Huang and Savkin [98]	Maximize coverage	Fixed	Not considered	2D	Local Search
Pinto and Lagorio [99]	Maximize efficiency	Fixed	Not considered	2D	CPLEX
ElSayed [102]	Maximize coverage and minimize energy cost	Energy consumption model	Considered	3D	GUROBI
ElSayed [11]	Minimize the number of charging stations	Energy consumption model	Considered	3D	GUROBI

.

Overall, the literature indicates that doubling the flight range reduces the number of charging stations by up to 77% while maintaining coverage [96]. More strict flight regulations lead to a 123% increase in the required number of charging stations to compensate for performance losses [11]. Two-dimensional modelling of UAV delivery overestimates the UAV performance due to unaccounted elevation and obstacles. In contrast, modelling in a 3D environment increases the number of required stations by 56% compared to a 2D solution [102]. Most reviewed studies assume fixed range and energy consumption [95,96,97,98,99]. However, it is worth noting that Hong [96] accounts for variable ranges according to the payload, and ElSayed [11], Mishra and Tiwari [100], and ElSayed [102] employ energy consumption models while accounting for external environmental factors.

A sustainable solution to UAVs’ energy use is attainable by integrating renewable energy sources into the charging infrastructure, although it is rarely explored in the literature [102]. Solar panels, for example, can be installed on rooftops and building facades, providing an eco-friendly energy source [102]. Wind turbines can be used in rural areas to supplement power generation. Hybrid systems combining solar and wind energy can enhance reliability, ensuring continuous power availability even in varying weather conditions [105]. However, challenges such as energy storage and variability in renewable energy generation must be addressed to ensure consistent charging station performance [105].

Several gaps remain despite the current progress in optimizing UAV charging stations. The real-time optimization of charging operations also needs improvement, with current systems lacking the ability to efficiently manage energy use, flight paths, and schedules under dynamic conditions [104]. Integrating mixed UAV fleets with varying battery capacities and sizes presents challenges in standardizing charging infrastructure. This requires optimizing the number of charging stations while ensuring compatibility across different UAV types [103]. Another gap is the insufficient focus on the UAV charging time; combined with charging station capacity constraints, this can result in queuing, potentially leading to significant increases in delivery time and costs [103]. Moreover, studies on alternative charging stations, such as streetlight-based charging, are limited, presenting another unexplored area.

2.7. Energy Consumption

The energy consumption of UAVs in parcel delivery is affected by several factors categorized into UAV-specific, mission-related, and external factors [106,107].

UAV-specific factors include weight, size, maximum takeoff weight (MTOW), the number and size of rotors, battery capacity, power efficiency, and flight speed [34,108]. For example, an unloaded octocopter is seven times heavier than a quadcopter, with six times higher energy consumption. However, their ranges are 4.2 km and 3.5 km, respectively [34]. While increasing battery capacity extends this range, there is a point where the additional battery mass becomes counterproductive, as shown in Figure 5 [34,71,109].

Another critical factor is the variable power efficiency that depends on the battery's state of charge (SoC) and UAV speed [34,107]. Defining trips by their duration, Rodrigues [35] claims that the total energy consumption during the trip remains constant as two opposite factors (trip distance and energy consumption per unit distance) are at play. Given a fixed time, a UAV travelling at a higher speed covers a greater distance while having lower energy consumption per kilometre, so the product between them remains constant. However, their conclusion is limited to the speed range investigated in their study (i.e., 4–12 m/s), and extrapolating these results reveals that energy consumption per km endlessly diminishes with an increasing speed.

This extrapolation is contradicted logically and through simulation results showing that energy consumption per unit distance increases with an increasing speed [34,108,110]. Energy consumption also non-linearly increases with speed, with higher speeds leading to disproportionate energy consumption [106]. Nevertheless, Rahmani [109] shows that energy consumption for hovering decreases as the speed increases from zero to the transition speed between hovering and cruising. From this transition speed onwards, energy consumption per distance increases, implying that this transition speed is optimal. The optimal speed was also found to increase with parcel weight, as heavier UAVs, particularly during takeoff and ascent, require more energy to generate the necessary thrust to overcome gravity. Cruising at the optimal speed reduced the total energy consumption by 27.2% [110].

Mission-related factors are flight segments, which include takeoff, landing, cruise, the number of parcels delivered per trip, and empty returns [106]. Several studies model energy consumption either by considering only the cruise segment [34,42,110] or all segments [35,108,109]. Overall, takeoff and landing count for 36% of the total energy consumption [35]; takeoff energy consumption is four times higher than landing [109]. Weight reduction resulting from the delivery of multiple parcels is often overlooked as a critical factor in estimating accurate energy consumption [34,42,110].

External factors are temperature, wind speed, and direction. Air density is directly related to atmospheric temperature and indirectly affects energy consumption. For instance, an increase in air density by 13.8% led to a rise in energy consumption by 3.6% [108]. Wind speed and direction significantly affect UAV energy consumption as flying against strong winds increases resistance, while tailwinds can reduce energy usage [107,108]. Parcel-related factors are the weight and size, as parcel size affects the drag ratio and can contribute to 9% of the total delivery energy [34]. Parcel weight changes the optimal velocity and can increase consumption by up to 100% [34,110]. Further operational challenges, such as infrastructure coordination, safety protocols, and integration with logistics systems, are highlighted in a recent study by Grote [111], emphasizing the need for holistic planning strategies in UAV logistics [111].

Table 8. The literature on energy consumption.

Studies	Factors Considered			UAV Type	Energy Consumption (Unit)
	UAV-Specific Factors	Mission-Related Factors	External Factors
Stolaroff [34]	Weight, size, MTOW, battery capacity, power efficiency, and flight speed	Only cruise, different consumption with and without payload	Wind speed and direction	Quadcopter	40 (J/m)
				Octocopter	340 (J/m)
Rodrigues [35]	Weight, battery capacity, and flight speed	All segments	Wind speed	Quadcopter	80 (J/m)
Zhao [110]	Weight, battery capacity, and flight speed	Only cruise, different consumption with and without payload	Wind speed	Hexacopter	65 (J/m)
Rahmani [109]	Weight, battery capacity, power efficiency, and flight speed	All segments	NA	Quadcopter	NA
Thibbotuwawa [108]	Weight, MTOW, battery capacity, and flight speed	All segments	Wind speed and direction	Multi-rotor	268 (J/s)
Dorling [71]	Weight, battery capacity, and flight speed	All segments	NA	Hexacopter	304 (J/s)

summarizes the literature on energy consumption, the factors considered, and energy consumption rates.

Zhang [106] thoroughly reviewed various approaches to model UAV energy consumption in parcel delivery, highlighting simple and complex methods. Integrated models, which rely on a single lift-to-drag ratio to estimate overall energy consumption, offer a broad assessment of UAV performance but often lack specificity regarding flight conditions or UAV configurations. In contrast, component-based models provide a more detailed analysis by breaking down flight dynamics into aerodynamic components such as drag, lift, and thrust. This method allows for the precise evaluation of energy consumption during different flight phases, including takeoff, cruising, and landing. Regression models, developed from field tests or simulations and calibrated with empirical data, establish predictive relationships between variables like weight, speed, and environmental conditions, delivering tailored energy consumption predictions for specific UAVs or missions.

Overall, the energy efficiency of UAVs diminishes with heavier parcels or in adverse weather conditions [107]. UAVs reduce energy consumption by up to 94% compared to diesel trucks and by 31% compared to electric vans per parcel delivered [34,35].

That said, the literature on UAV energy consumption indicates several significant gaps. First, many existing studies fail to account for all stages of flight, including takeoff, landing, and cruise, which can impact overall energy consumption. Additionally, the relationship between UAV weight, speed, and energy consumption is often oversimplified despite evidence showing a non-linear relationship [108]. The effects of environmental factors, such as wind and temperature, are also underexplored, particularly in adverse conditions like strong winds or extreme temperatures, which can drastically affect performance [107]. Furthermore, experimental research is limited, and when conducted, it often shows significant discrepancies compared to theoretical models [106]. The impact of parcel weight on energy consumption, especially when comparing inbound and outbound trips, lacks clear identification [110]. These gaps highlight the need for more comprehensive experimental studies, particularly under real-world conditions, to ensure UAVs can be viable for critical delivery applications, especially in adverse weather.

3. Saturated Knowledge for UAV Last-Mile Delivery Research

This section synthesizes the saturated knowledge of UAV last-mile delivery research across the seven investigated domains. Saturated knowledge refers to recurring insights and observations highlighted in the reviewed studies and derived from their collective analysis.

Environmental performance: UAVs consistently produce lower GHG emissions compared to truck delivery. These emissions range from 1.9 to 79 g Co2eq/km and from 0.6 to 1786 g Co2eq per parcel, significantly outperforming truck delivery in various studies [34,36,37,38]. The current literature extensively examines UAV-related GHG emissions under different operational conditions, including regulatory constraints, demand density, different electricity sources, and payload capacity [31,32,35,39,40].

Economic impacts: UAVs offer substantial cost savings, reducing delivery costs by up to 96.5% compared to truck delivery, with per-parcel costs ranging from USD 0.03 to USD 0.67. Additionally, truck-assisted UAV systems reduce the number of required trucks by 27% and cut costs by 48% [16,41].

Social impacts: The social acceptance of UAVs varies significantly across countries and socio-economic and demographic profiles [29,54]. Most studies rely on surveys to examine public concerns regarding safety, privacy, and acceptance [32,53,54,55].

Policy and regulations: Strict regulations significantly influence UAV operations, increasing costs by up to 10%, reducing emissions by up to 400%, and limiting operational range by 75%, especially in urban contexts [11,38]. The current literature on UAV last-mile delivery thoroughly focuses on regulatory frameworks, operational limitations, classifications, and ethical considerations to address these challenges [24,60,61].

Routing and scheduling: The optimization of UAV last-mile delivery routing and scheduling is widely explored, with studies addressing both standalone UAVs and truck-assisted UAVs for single and multiple parcel deliveries [67,75,87]. However, most models are limited to 2D frameworks without considering the complexities of 3D environments [86,87].

Charging infrastructure: Multiple studies focus on optimizing the location of stationary and mobile charging stations while considering the UAV’s operational range, regulation constraints, and environmental conditions [95,96,97,98,99,100]. It is reported that doubling the flight range reduces the number of stations by up to 77%, whereas strict regulations and 3D modelling increase station numbers by 123% and 56%, respectively [11,96,102].

Energy consumption: UAVs in last-mile delivery consume 94% less energy than diesel trucks and 31% less than e-trucks [34,35]. The current literature examines energy consumption factors, including flight speeds, parcel weight, flight segments, wind conditions, and battery capacity [34,71,109].

4. Knowledge Gaps, Potential Interventions, and Policy Implications

This study identifies critical research gaps for UAV research in last-mile delivery across seven domains, presenting opportunities to enhance the feasibility, sustainability, and efficiency of UAV operations.

Environmental: There is a need for sensitivity analyses to identify the most critical factors influencing UAVs’ environmental performance. Additionally, cradle-to-grave lifecycle assessments are essential to compare UAV deliveries with truck-assisted UAVs, diesel trucks, and e-trucks.

Economic: Conducting sensitivity analyses while considering total system costs (fixed and operational) is essential. Furthermore, most studies overlook the costs associated with charging infrastructure, which should be integrated for a comprehensive economic evaluation [17,46].

Social impacts: Research on social impacts requires experiments involving human interaction to better assess public acceptance and concerns, particularly noise, size, and trajectory [59]. Discrepancies in social impact assessments highlight the need for standardized methods to improve reliability and comparability across studies [29].

Policy and regulations: Current regulations lack frameworks for autonomous UAV operations, highlighting the need for comprehensive guidelines to ensure safe and efficient operations [11]. Policies and regulations should also address interactions between autonomous UAVs and road users at low altitudes.

Routing and scheduling: Most routing and scheduling models require experimental validation under diverse operational scenarios [7]. Developing optimization models for swarm UAV routing is crucial to improving operational efficiency. Integrating charging station availability into these models is also necessary for more realistic and practical outcomes.

Charging infrastructure: There is a lack of real-time optimizations of charging allocation while considering energy management, flight paths, and dynamic charging scheduling [104]. There is also a need to standardize charging stations to accommodate different UAV types and battery capacities [103]. Investigating alternative charging stations, such as streetlight-based charging, is also recommended.

Energy consumption: Current models oversimplify the non-linear relationship between UAV weight, speed, and energy consumption, highlighting the need for more accurate models [108]. Further research should examine the influence of temperature and wind direction on energy consumption [107]. Experimental validation under varying operational conditions is also essential to enhance prediction accuracy [106].

Considering the intersectionality of multiple research domains, we developed an intersectionality matrix (Figure 6). The matrix was developed to examine the intersection, or the lack thereof, between two knowledge domains in UAV last-mile delivery research.

The matrix highlights several under-researched areas. There is a notable gap in the research addressing the integration of charging infrastructure allocation with other domains, including social acceptance, environmental performance, and energy consumption. Furthermore, the intersection between routing/scheduling optimization under various regulations is under-researched. In contrast, the intersections between economic and environmental and charging allocation with energy consumption are very well-researched domains.

Based on the synthesized saturated knowledge and knowledge gaps, this study proposes future interventions to address UAV last-mile delivery knowledge gaps, detailed in Figure 7. The core aspect of these interventions is conducting experimental research. The proposed interventions are linked to targeted research gaps, informing future studies.

5. Conclusions

In conclusion, this study provides a comprehensive review that synthesizes the state of the art on implementing UAVs in last-mile delivery, examining seven key research domains: environmental performance, economic impacts, social impacts, policy and regulations, routing and scheduling, charging infrastructure, and energy consumption.

A review of 98 relevant studies highlights the potential of UAVs to substantially reduce GHG emissions (up to 71%) and operational costs (up to 97%) compared to truck delivery methods. However, strict regulations, on average, increase costs by up to 10%, reduce emissions by up to 400%, and limit operational range by 75%. The findings also indicate that integrating UAVs with renewable energy sources could achieve zero net emissions for last-mile delivery. Furthermore, the social impacts of UAVs vary considerably across countries, socio-economic demographics, and geography (urban vs. rural)

Despite these promising advancements, there exist several critical knowledge gaps. Environmentally, sensitivity analyses are needed to identify the most influential factors affecting environmental performance and cradle-to-grave lifecycle assessments to compare UAVs with alternative delivery methods. Economically, comprehensive evaluations must include the cost associated with charging infrastructure. Socially, experiments incorporating human interactions are necessary to better evaluate public acceptance and concerns. In terms of policy, frameworks for autonomous UAV operations must be developed to ensure safe and efficient use. Routing and scheduling models require experimental validation under varying scenarios, and integrating charging station availability into these models is crucial. From a charging infrastructure perspective, the real-time optimization of the charging schedule while considering energy management, flight paths, and schedules under dynamic conditions is essential. Standardizing charging stations to accommodate UAVs with different battery capacities and sizes is also necessary. Additionally, energy consumption models must reflect the non-linear relationships between UAV weight, speed, and energy consumption, along with the impacts of external factors such as temperature and wind directions.

This review offers future interventions to address identified knowledge gaps and realize the full potential of UAVs in last-mile delivery. Future research should focus on conducting experimental validations under varying conditions and considering real-world factors (i.e., wind speed, direction, and moving obstacles). Additionally, exploring the social impacts and human interactions, as well as refining policy frameworks, will be essential for the widespread adoption of UAVs. Future research should also investigate how optimizing UAV system configuration (i.e., component sizing and charging infrastructure design) impacts economic performance, routing, scheduling, and energy consumption. Covering these areas would enable the academic community, industry stakeholders, and policymakers to collaboratively advance the implementation of UAVs into sustainable and efficient last-mile delivery systems.

Finally, we acknowledge that this review has two primary limitations. First, the scope of the review focuses only on studies published within the past ten years, which may exclude relevant earlier research. Second, the rapid evolution of UAV technology may render some findings outdated, highlighting the need for ongoing updates to reflect the latest advancements. Consequently, some of the identified knowledge gaps might not be addressed extensively enough.