Brasil
Ciência Rural
Tabla de contenido Anterior Actual Siguiente
Tabla de contenido
RURAL ENGENEERING • Cienc. Rural 56 (09) • 2026 • https://doi.org/10.1590/0103-8478cr20250501 linkcopy

Open-access Planning and optimization of spraying missions for agricultural drones

Planejamento e otimização de missões de pulverização para drones agrícolas

AuthorshipSCIMAGO INSTITUTIONS RANKINGS

ABSTRACT:

This study presents an optimized mission planning strategy for agricultural unmanned aerial vehicles (UAVs) that integrates battery endurance constraints into the coverage path planning (CPP) problem. The primary objective was to minimize total mission time by optimizing the sweep direction (θ) and automatically segmenting the path into executable sorties (m) based on battery capacity. The methodology was validated through simulations and a real-world case study of a 7.02 ha sugarcane field using parameters from a DJI AGRAS T10 drone (swath width of 5 m and flight speed of 5 m/s). The proposed approach was compared against classical methods, specifically those of TORRES et al. (2016) and CHOSET & PIGNON (1998). Results indicated reductions of 2.65% and 8.2% in total mission time, respectively, achieved by reducing unproductive turns and optimizing battery usage. This efficiency gain directly extends the operational window for spraying, enabling larger areas to be treated under ideal weather conditions. The findings demonstrated that integrating operational endurance into offline planning improves the reliability and logistics of drone-based precision agriculture applications.

Key words:
unmanned aerial vehicles (UAV); aerial application; optimization models; site-specific management

RESUMO:

Este estudo apresenta uma estratégia de planejamento de missão otimizada para Veículos Aéreos Não Tripulados (VANTs) agrícolas que integra restrições de autonomia de bateria ao problema de planejamento de caminhos de cobertura (CPP). O objetivo do estudo foi minimizar o tempo total da missão por meio da otimização do ângulo de varredura (θ) e da segmentação automática do percurso em surtidas (m) compatíveis com a capacidade da bateria. A metodologia foi validada por meio de simulações e um estudo de caso real em uma área de 7,02 ha de cana-de-açúcar, utilizando parâmetros de um drone DJI AGRAS T10 (faixa de aplicação de 5 m e velocidade de 5 m/s). A abordagem proposta foi comparada com métodos clássicos, especificamente os de TORRES et al. (2016) e CHOSET & PIGNON (1998). Os resultados indicaram uma redução no tempo total de missão de 2,65% e 8,2%, respectivamente, ao reduzir manobras improdutivas e otimizar o uso da bateria. Além disso, esse ganho de eficiência amplia diretamente a janela operacional de pulverização, permitindo que áreas maiores sejam tratadas sob condições climáticas ideais. Portanto, os achados demonstram que a integração de restrições energéticas no planejamento offline melhora a confiabilidade e a logística das aplicações aéreas com drones na agricultura de precisão.

Palavras-chave:
veículos aéreos não tipulados (VANT); tecnologia de aplicação; modelos de otimização; manejo sítio-específico

INTRODUCTION

The demands of global food security and sustainable production have demanded the integration of automation and robotics technologies in agricultural operations. In this context, agricultural drones have been introduced as a vital technology applied to precision spraying (CHEN et al., 2021) that enable high-resolution aerial herbicide application, reducing environmental impact compared to traditional methods. Consequently, the adoption of agricultural drones has significantly enhanced various precision agriculture (PA) practices, including crop mapping, real-time monitoring, automated spraying, and yield prediction (GUEBSI et al., 2024). Nevertheless, despite their widespread use and numerous advantages, agricultural drones still face several technical and operational challenges that must be overcome to fully realize their potential and drive further advancements in the field.

Agricultural Unmanned Aerial Vehicles (UAVs) for spraying are specialized aircraft configured to carry liquid payloads, enabling the localized application of pesticides and fertilizers with reduced environmental drift and optimized chemical use (HAFEEZ et al., 2023). These UAVs are typically equipped with advanced devices and sensors commonly used in precision agriculture, including Real-Time Kinematic (RTK) GPS, multispectral cameras, temperature sensors, and liquid tanks and sprayers. By collecting and analyzing real-time data, agricultural drones support resource management, enhance operational efficiency, and improve crop health and productivity (MOGILI & DEEPAK, 2018). However, the effectiveness of these advanced hardware systems in the field is heavily dependent on the underlying Coverage Path Planning (CPP) algorithms, which determine the efficiency of the flight aerial application. Coverage Path Planning (CPP) involves generating an aerial application that enables a drone to efficiently cover a designated area. Effective CPP must account for the dynamic constraints inherent to the drone’s coverage path. Numerous studies have explored the fundamental concepts and methodologies employed in this field (CABREIRA et al., 2019). A central focus in CPP research is the optimization of flight paths, with the primary objectives being reducing operational endurance and minimizing total path duration (DI FRANCO & BUTTAZZO, 2016).

To this end, TORRES et al. (2016) developed a coverage path that reduces battery consumption by minimizing the number of turns required to cover complex regions, based on the geometry of the coverage areas. The coverage path planning is optimized subject to a limited onboard operational endurance (BOUZID et al., 2017). The optimization of the visiting order and the optimization of the flight lines orientation were proposed for the coverage path planning of disjoint areas (VASQUEZ-GOMEZ et al., 2018). Global optimization using a genetic algorithm for coverage path planning was reported in (YUAN et al., 2022). Multiple drones were used to cover large areas in agricultural applications as an alternative to the coverage path-optimization approach (MUKHAMEDIEV et al., 2023). LARA-MOLINA (2025) optimizes the drone spraying mission to cover weed-infested areas of a sugarcane crop. Moreover, coverage path methods have been used in drone mission planning for crop monitoring (MUKHAMEDIEV et al., 2023; MANSUR et al., 2025) and precision agriculture for aerial spraying (HUANG et al., 2024; LARA-MOLINA, 2025).

Several methods have been applied to optimize the coverage path planning of agricultural drones. TEVYASHOV et al. (2021) applied iterative minimization of the maximum time needed to cover the assigned sub-areas. MUKHAMEDIEV et al. (2023) applied genetic algorithms to optimize the coverage path of multiple drones. LARA-MOLINA (2025) also used genetic algorithms to optimize coverage of weed-infested sugarcane crops using the traveling salesman problem. YANG et al. (2025) optimized the region segmentation and path orientation using eight-directional A* search combined with polyline simplification, arc fitting, Chaikin subdivision, and B-spline smoothing. A robust modified simulated annealing (MSA) algorithm is introduced to determine the shortest path coverage route (FAHAD et al., 2025). Considering the flight-time limitations of agricultural drones, such as battery endurance, planning missions is crucial to maximizing flight time in large areas. The studies reported by MUKHAMEDIEV et al. (2023) and LARA-MOLINA (2025) optimized the coverage paths of agricultural drones under dynamic constraints.

Nevertheless, additional developments are required to include optimal approaches into comprehensive mission planning for agricultural drones. Specifically, further developments in mission planning for agricultural drones are required to optimize spraying operations. This included defining coverage mission parameters, such as the sweep angle, while accounting for battery endurance to minimize operational time.

The efficiency of aerial spraying is limited by a brief period when wind, temperature, and humidity are ideal for application. Therefore, the development of path optimization is a crucial issue for farm management rather than just a technical challenge. Spraying UAVs can treat larger areas when weather conditions are optimal, reducing the risk of spray drift by minimizing mission time.

The present study proposed a novel approach to optimize mission planning for spraying drones while accounting for battery endurance constraints. The proposed method aimed to design an offline coverage path planning strategy that includes the following contributions: i) A comprehensive framework is developed for a mission planning strategy to optimize drone spraying operation in agricultural missions. ii) The present approach integrates the operation requirements of a typical spraying operation, such as spraying area (target area), and battery endurance constraints to optimize drone operation. Thus, the sweep direction is optimized based on dynamic specifications and battery constraints. iii) The complete mission is split into several submissions that will depend on the agricultural drone’s battery endurance and operational conditions; moreover, the proposed approach is compared against classical methods proposed by TORRES et al. (2016) and CHOSET & PIGNON (1998).

This research developed an optimized offline mission planning strategy to enhance the operational performance of agricultural spraying drones. By determining the optimal sweep direction to cover a target area, the proposed approach minimizes total flight time and energy consumption, addressing critical constraints such as battery endurance and field coverage efficiency.

MATERIALS AND METHODS

An agricultural spraying operation uses unmanned aerial vehicles (UAVs) to distribute agrochemicals such as herbicides, pesticides, or fertilizers across crop areas. The present approach aimed to plan optimally for a spraying operation using an agricultural drone.

The approach proposed in the present contribution is presented in the flow diagram in figure 1. This method is composed of three main stages: i) data inputs, ii) mission optimization and segmentation, and iii) flight plan specification.

thumbnail of Figure 1
Figure 1
Flowchart of the mission planning strategy. The process considers input data from the drone’s operational parameters and mission optimization, and produces outputs for the flight plan.

Initially, the required input data are agricultural field imagery and the drone’s operational settings. Agricultural field imagery can be drone-mapped crop images or satellite images; these images should be georeferenced to specify the target area for spraying. Moreover, the drone’s operational settings include speed, time, and battery endurance; these specifications vary by drone model.

The proposed method consists of two main stages: i) determining the sweep direction to minimize the mission execution time, and ii) segmentation of the mission depending on the battery’s endurance. Finally, the flight plan and its specifications are determined as the output approach. These data can be used to implement the mission on the agricultural drone.

The mathematical model for mission planning is based on a combination of classical kinematics and geometric coverage path formulations frequently adopted in UAV research (CABREIRA et al., 2019). The total mission time is calculated as the sum of productive spraying time and unproductive turnaround intervals, following the fundamental relation t = d/v, where d represents the segment length and v the constant flight speed.

Data inputs and model assumptions

To ensure the mathematical coherence and reproducibility of the mission planning strategy, the following assumptions were adopted: i) Flight dynamics: The UAV maintains a constant flight speed (v) and constant altitude during the productive stages of the mission. ii) Environmental conditions: The model assumes ideal weather conditions with negligible wind influence on the aerial application and spray pattern. iii) Field geometry: The target areas are considered convex polygons free of internal obstacles. iv) Spray parameters: The application rate and flow control are assumed to be uniform across the entire coverage path. The operational parameters used for validation were based on the DJI AGRAS T10 platform. Technical specifications include a 10-liter tank capacity, a flight speed of 5 m/s, and a 5-meter swath width (δr). The spraying system features four high-pressure nozzles (typically XR11001VS), ensuring a consistent droplet spectrum for precision applications.

In the framework of this study, the drone configuration is represented by the state vector q = (pd , θ), where pd = (x d , y d ) denotes the position coordinates in the north-east-down (NED) reference frame, and θ indicates the heading angle relative to the east axis. ω and v correspond to the drone’s angular velocity (i.e., yaw rate) and its translational velocity along a linear aerial application, respectively (Figure 2). Additionally, the coverage path is defined based on the following operational settings of the drone (Figure 3): row spacing (δ r ), home position (p H ), and battery endurance (t a ).

thumbnail of Figure 2
Figure 2
Schematic representation of the drone configuration in the 2D operational plane. The state is defined by its Cartesian position (xd and yd) and heading orientation (θ) relative to the global inertial reference system.

thumbnail of Figure 3
Figure 3
Coverage path planning with battery endurance constraint over the target area M: path with two soties (m = 1 and 2), sweep direction (θ), home position pH, row spacing (δr), and eighth row lines Lj (j = 1, 2, ..., 5).

In typical spraying missions, the target area to be covered by the drone can be effectively represented by a convex polygon characterized by two main assumptions: (i) the area is usually delineated using a relatively small set of vertices, and (ii) there are no obstacles exceeding a predefined height within the region (DI FRANCO & BUTTAZZO, 2016). Under these considerations, the coverage region is modeled as a convex polygon M comprising n v vertices vi , such that: M = v1, v2, …, vn v , where each vertex vi = (x i , y i ) belongs to R2 for i = 1, …, n v .

The target area for the spraying operation was defined using high-resolution satellite imagery from ESRI (ESRI et al., 2021), and the field boundaries were manually delineated in QGIS. This resulting target area was then exported as a GeoJSON file and imported into Python using the GeoPandas library for further processing and visualization.

Mission optimization

The mission planning strategy consists of two steps: i) the minimization of mission time execution (section “Minimization of Mission Time”) that aims to determine the optimal sweep direction to minimize the missions’ time execution and ii) the segmentation of the target area (section “Segmentation of the Target Area”) that decomposes the entire target area into several polygons that can be covered by the drone with a single charge of battery.

Minimization of mission time

Given that different sweep angles (θ) yield varying numbers of turns and path lengths depending on the geometry of the polygon, it becomes essential to select the sweep direction that minimizes the overall coverage cost (Figure 3). This cost, denoted by c(θ) (Eq. 2.1), is formulated as a function of the total mission time, the number of directional changes (turns), and the kinematic parameters of the drone, including its linear velocity (v) and angular velocity (ω).

The expression used to evaluate the path execution time corresponds to the total time required to traverse all waypoints, as defined in Eq. 2.1.

Ttotal=cθ=∑ic(pi,pi+1)(2.1)

Let p i and p i+1 denote two consecutive waypoints along the j-th coverage line. The time required for the drone to reorient and travel between these two waypoints is given by the cost function of Eq. 2.2:

cpi,pi+1=θΔωpi+1-piv(2.2)

where θ ∆ represents the change in heading angle between the segments, v is the drone’s linear velocity, and ω is its angular velocity.

To find θ * ∈ [0, 2π), a set of candidate angles is evaluated; for every evaluated θ * angle, a path is computed, and its associated cost is calculated. The angle that yields the minimum cost is then selected as the optimal sweep direction for that polygon according to Eq. 2.3.

Therefore, the following conditions are considered in the formulation of the present optimization problem:

A target area M ⊂ R2

A fixed number J of parallel sweep lines {L 1, L 2,. …, L j , …, L J }

A home/refueling station at position pH ∈ R2

A coverage footprint width δ r , determining the spacing between sweep lines

A maximum flight time (fuel constraint) t a

The operational capacity (C) is considered to evaluate the efficiency of the spraying operation. This metric is calculated as C = M/Ttotal, where M is the target area (Figure 3), and Ttotal is the total mission time (according to Eq. 2.1).

The design variable in this optimization problem is the angle between the sweep lines and the x-axis. This angle, denoted by θ, determines the orientation of all coverage lines L j and; consequently, the corresponding set of waypoints. The variable is constrained within the interval θ ∈ [0, 2π).

Each sweep line L j (θ) intersects the polygon and is discretized into two waypoints that intercept the polygon L j (θ) = {p i , p i + 1}. Each waypoint p i ∈ R2 for i = 1, 2... is inside the polygon.

The objective is to identify the optimal sweep direction, denoted by θ *, for a given target area M, as this parameter significantly influences the efficiency of the coverage path within the designated area (Figure 3). The sweep direction establishes the orientation of a set of parallel lines. The systematic back-and-forth coverage path pattern ensures complete coverage of the target region. Thus, minimize the total mission time of Eq. 2.3, considering that the drone must return to p H whenever the accumulated path time exceeds t a:

minθ∑m=1M∑i∈Smθcpi,pi+1(2.3)

where: - Sm (θ) is the ordered segment path of sortie m under sweep angle θ, - M is the number of sorties required (i.e., partial tours limited by t a ). The present optimization problem is subject to the following constraints:

Sweep lines ordered: L 1(θ) → L 2(θ) →...→ L J (θ). Moreover, each waypoint covered exactly once: Um=1MSmθ=Uj=1JLj .

The total time for each sortie is given by Eq. 2.4.

∑i∊Smcpi,pi+1+cpstartm,pH+cpendm,pH≤ta(2.4)

with: pstart1=pendM=pH

Segmentation of the Target Area

Let M ⊂ R2 be a convex polygonal region to be covered. Let θ ∈ [0, π) be the fixed sweep direction (angle) defining a family of parallel sweep lines over M. The full coverage path is partitioned into a sequence of M sorties, such that: Pθ=Um=1MSmθwhere Pθ is the complete coverage path under sweep angle θ, Sm (θ) = {p (m) }n m is the m-th sortie path, a sequence of ordered segments that begin and end at the refueling station (base), and M is the total number of sorties required to cover the full region under the endurance constraint. Consequently, the original region M is partitioned into M sub-regions M1, M2, …, MM , as shown in Eq. 2.5.

M=Um Mm(θ) (2.5)

where Mm (θ) ⊂ M is the area covered during sortie m and depends on the sweep direction θ. Each Mm (θ) corresponds to the spatial footprint of the path Sm (θ).

Mission Planning outputs

The mission planning algorithm outputs a set of parameters that define the drone’s flight behavior to ensure complete coverage of a convex region under an endurance constraint. These outputs are defined as follows:

M = {M1, M2, …, MM }: A set of M sub-polygons such that each Pm ⊆ M corresponds to a region covered in sortie m, where M is the total area to be covered. Each Mm is defined by its ordered vertex list in Cartesian coordinates.

θ ∈ [0, 2π): The global sweep angle, standard to all sub-polygons, defines the orientation of the parallel coverage lines concerning a fixed axis (e.g., the x-axis). This angle is a design variable optimized to minimize the required sorties M or the total coverage time.

v ∈ R+: Constant velocity of the drone during the coverage operation. This value must be set according to the spraying or sensing requirements and hardware constraints.

pH ∈ R2: The Cartesian coordinate of the base station (initial and final point of all sorties). Each sortie is assumed to start and end at pH .

These outputs serve as the basis for autonomous mission execution by the agricultural drone and are typically exported in a mission planning file format (e.g., JSON, CSV, or proprietary formats). Moreover, the spraying mission is entirely defined by specifying the following parameters:

δr ∈ R+: Constant coverage footprint that determines the spacing between sweep lines during the coverage operation. This parameter depends on the width of the sprayer nozzle mounted on the drone and ensures that the entire area is fully covered, with no gaps or excessive overlap.

h ∈ R+: Constant altitude of the drone during the coverage operation. This parameter is chosen based on terrain characteristics and operational constraints. The value of h influences the effective coverage footprint δ r , flight safety, and precision.

Spray amount and flow rate: These parameters are determined by the crop and the type of herbicide being applied. Proper calibration is essential to ensure the correct dosage per unit area, which depends on plant density, growth stage, and product specifications.

RESULTS AND DISCUSSION

The present application considers the DJI AGRAS T10 drone. Accordingly, the following fixed parameters were adopted for the case study to optimize the mission planning based on the proposed method described in Section “Materials and Methods”: constant velocity v = 7 m/s, coverage footprint spacing δ r = 7 m, maximum angular velocity ω = 45◦/s, and autonomy time per sortie t a = 420 s (7 minutes). The proposed algorithm was implemented in Python, and all numerical simulations were performed on a system running Ubuntu 20.04 with an Intel Core i7 processor.

This section considers the solution to the minimization of the mission time in section “Minimization of Mission Time”, the segmentation of the target area in section “Segmentation of the Target Area”, and the application of the proposed approach to a real drone spraying mission in section “Materials and Methods”.

Minimization of mission time

The optimization problem presented in Section “Minimization of Mission Time”, aiming to minimize the total mission time by selecting the optimal sweep direction defined in Eq. 2.2, was numerically solved. For the case study, eight distinct targer areas were considered, as illustrated in figure 4. These polygons included shapes with varying complexity: a triangle (three vertices, Figure 4A), quadrilateral (four vertices, Figure 4B), pentagon (five vertices, Figure 4C), hexagon (six vertices, Figure 4D), heptagon (seven vertices, Figure 4E), octagon (eight vertices, Figure 4F), nonagon (nine vertices, Figure 4G), and a complex shape with thirty-six vertices (Figure 4H).

thumbnail of Figure 4
Figure 4
Representation of the eight distinct polygonal target areas used to evaluate the mission planning strategy (Polygons 1-8). These areas define the operational boundaries for the drone’s coverage path planning.

The coverage paths obtained for the eight considered polygons are presented in figure 4. Beyond the geometric variety, these results demonstrated the model’s ability to maintain high coverage efficiency regardless of the number of vertices. The paths illustrate the systematic interruption and resumption of the mission; when the battery endurance limit (t a ) is reached, the UAV automatically returns to the home station for charging; and subsequently, returns to the exact coordinates where the spraying was halted. This behavior is critical for operational reliability, ensuring that no areas are left untreated and avoiding excessive overlaps that could lead to phytotoxicity or chemical waste. Furthermore, the alignment of the sweep direction (θ) with the longest edge of the polygons in most cases confirmed that the optimization successfully minimizes the number of energy-demanding turns, even in more complex shapes like the 36-vertex polygon (Figure 4H).

The optimized sweep angles aligned with the longest axis of each polygon, effectively minimizing the number of turns and maximizing coverage efficiency per sortie. This behavior confirmed the influence of geometric orientation on the overall mission time. The number of sorties (M) increased proportionally with the polygon’s perimeter and area, subject to the drone’s the maximum flight endurance.

The proposed approach was compared with the coverage path methods proposed by TORRES et al. (2016) and CHOSET & PIGNON (1998). The coverage paths (CPs) obtained using these methods were examined for the polygon with eight vertices, as shown in figure 4G. The resulting CPs were evaluated under the total time constraint defined in Eq. 2.5, ensuring a fair comparison among the three CP methods considered.

As illustrated in figure 5, the differences in coverage path definitions, specifically, the selection of the initial waypoint, final waypoint, and sweep angle orientation, directly influence the number of required sorties. The method of CHOSET & PIGNON (1998) adopts a canonical back-and-forth sweeping pattern aligned with the principal axis of the polygon’s bounding box, which may not always be optimal when endurance constraints are present. The approach by TORRES et al. (2016) employs line-sweep direction optimization based on the longest-edge alignment, providing improved performance but still disregarding endurance limitations in its planning phase.

thumbnail of Figure 5
Figure 5
Comparison of coverage paths (CP) generated by different algorithms for a sample target area: (A) the proposed mission planning strategy; (B) the method based on TORRES et al. (2016); and (C) the Boustrophedon decomposition by CHOSET & PIGNON (1998). The geometric variation between paths (A), (B), and (C) illustrates the impact of trajectory optimization on field efficiency.

In contrast, the proposed method integrates the endurance constraint into the sweep angle optimization, effectively minimizing the number of sorties by selecting a sweep direction that reduces unnecessary returns to the home position. As a result, the proposed approach yields a path with fewer sorties and reduced total mission time compared to the methods of TORRES et al. (2016) and CHOSET & PIGNON (1998).

Table 1 summarizes the number of sorties, path length, and mission time obtained with each method. For example, for polygon 1, the proposed method achieved a reduction of approximately 2.65% in total mission time compared to TORRES et al. (2016) and 8.2% in total mission time compared to CHOSET & PIGNON (1998), demonstrating the effectiveness of explicitly considering endurance constraints in the mission planning phase. The behavior observed for polygon 1 is also observed for all the polygons analyzed in table 1; i.e., the method proposed helps reduce the mission’s flight time. To evaluate the practical efficiency of the proposed method, the results were converted into operational capacity (C), expressed in hectares per hour (ha h-1) (Table 1). For the 7.03 ha (polygon 8 (Figure 4H) of Table 1) case study, the proposed optimization increased the operational capacity from 14.152 ha h-1 to 15.632 ha h-1 compared to the CHOSET & PIGNON (1998) method.

Thumbnail
Table 1
Performance metrics for the proposed coverage path (CP) strategy across eight distinct target areas. The evaluation includes total path length (m), execution time (s), and the resulting effective operational capacity (C, ha h-1) for each polygon geometry.

In practical terms, this implies a reduction in mission time and the number of recharges required to complete the operation. Additionally, reducing the number of sorties decreases the total mission time. It minimizes the operational endurance required for take-off and landing cycles, which is critical for increasing operational efficiency in practical field deployment. For large-scale producers, where hundreds of hectares must be treated within narrow environmental windows, an 8% increase in efficiency can yield an additional 4 to 6 hectares treated per day per drone. Furthermore, by optimizing path continuity, the model ensures the drone returns to the base only when the battery is near its safe discharge limit, maximizing the effective spraying time and reducing the total number of non-productive battery swap cycles over the course of a growing season.

Although, the reduction in mission time seems small geometrically, it is operationally significant. This efficiency allows for greater field coverage during the limited operational window for spraying before weather conditions change. Completing more sorties early in the day is crucial for timely pest and disease control, especially when sudden weather shifts might otherwise force the suspension of operations.

Segmentation of the target area

This section analyzes the second stage of the mission planning strategy described in figure 1, which consists of segmenting the target to split the several sorties that define the entire mission.

The proposed method’s coverage path (CP) was examined for the polygon with eight vertices, as shown in figure 5A. The resulting coverage paths were evaluated and divided into six sorties, as shown in figure 6, depending on the battery endurance. These results indicate that the entire mission path can be divided into six single missions, each defined by a sub-polygon.

thumbnail of Figure 6
Figure 6
Decomposition and path generation for a complex boundary: segmentation of the target area (Polygon 7 from Figure 4) into convex sub-regions and the resulting optimized coverage paths (CP).

This segmentation process ensures that each sortie remains within the maximum allowable flight time, thus preventing mid-mission battery depletion and guaranteeing operational safety. Furthermore, by structuring the mission as a sequence of sorties, the proposed approach provides a structured and systematic execution of the spraying mission.

This segmentation process was derived from the optimization process that seeks to balance path continuity and minimize unnecessary returns to the base. This strategic partitioning reduces overlap between adjacent sorties and aligns the sweep lines across sorties to avoid coverage gaps or redundancies.

Additionally, this modular mission structure simplifies mission monitoring and management. For instance, when unforeseen obstacles or environmental conditions interrupt the mission, the operator can resume the mission from the last completed sortie without replanning the entire operation.

Furthermore, the proposed method dynamically segments the coverage task into sub-polygons (implicitly defined by the sortie path planning), ensuring full-area coverage while adhering to endurance constraints. The total mission path and number of sorties for each polygon are summarized in table 2. These results considered the vertices of the obtained sub-polygons M with their corresponding area; moreover, the sweep direction (θ) was obtained in the previous procedure (section “Minimization of mission time”).

Thumbnail
Table 2
Coverage path outputs for mission planning presented in figure 6. The data includes the segments of the target area (Mi), corresponding area (A), sweep direction (θ), and the fligth time (c).

Application to a spraying mission - case study

The proposed mission-planning strategy, as outlined in figure 1, was applied to optimize a real spraying task using the DJI AGRAS T10 drone. This case study aimed to evaluated the practical applicability of the method in an agricultural context.

For simulation and validation, operational parameters based on the standard configuration of the DJI AGRAS T10 drone for spraying medium-sized crops were adopted descrided in section “Data Inputs and Model Assumptions”. Our model maintains constant flight speed (v) and swath width (δr), which are essential for uniform dosing and preventing over-application.

The target area (Figure 7A) corresponds to a sugarcane plantation in Planura-MG, Brazil, using QGIS, resulting in a rectangular polygon representing the crop area of 7.02 ha. At the present stage of the research, no operational specifications for spray application, such as application rate or spray volume, were included in planning the UAV’s mission; i.e., the study considers only aspects related to the path.

thumbnail of Figure 7
Figure 7
Strategic mission decomposition for large-scale operations: (A) 7.02 ha operational area; (B) total theoretical path; and (C) actual mission execution split into multiple sorties.

Subsequently, the total coverage path was computed by solving the optimization problem described in Section “Minimization of Mission Time”. The optimal sweep direction was determined to be θ = 335◦, which minimized the total mission time. The solution required dividing the mission into four sorties (m = 4), each corresponding to a complete drone battery cycle. This result indicated that, given the payload constraints and the DJI AGRAS T10’s battery autonomy, the spraying task for the specified field must be executed in four separate flights.

The proposed methodology effectively determined the optimal sweep direction and the number of sorties needed to cover the target area under realistic operational constraints. These characteristics reinforce its applicability to real-world precision agriculture scenarios, ensuring efficient path planning while respecting battery and coverage limitations. Integrating battery endurance (ta) into mission planning optimizes battery use and logistics. The proposed method saves energy by reducing unnecessary turns, which is vital for large-scale operations in Brazil, where charging infrastructure and battery cycles limit daily productivity. In this case study, the proposed optimization increased operational capacity from 14.152 ha h-1 to 15.632 ha h-1 compared with the CHOSET & PIGNON (1998) method.

Finally, figure. 7c shows the segmentation of the entire target area based on the number of sorties. The resulting sub-polygons represent distinct subareas of the mission, each corresponding to a single sortie. These segmented areas can be programmed into the drone’s control system using the mission parameters defined for the spraying task: row spacing (δ r ), drone velocity (v), flight altitude (h), and the optimized sweep direction (θ). This segmentation ensures that each sortie complies with the drone’s operational limitations while maximizing coverage efficiency.

Although, this study centers on aerial application and battery optimization, these factors are fundamental to the technology’s application. By automating mission resumption, the strategy minimizes human errors that cause non-uniformity in manual operations. While a gain may seem small in a single flight, it is highly relevant for large-scale operations. Future research will use water-sensitive papers and tracers to measure deposition and drift along these optimized paths.

Advantages and limitations

The proposed approach presents several advantages for mission planning in agricultural drone applications. The robustness of the proposed method was tested across eight diverse geometric configurations, ensuring its applicability to various field shapes. While the current case study provides a successful proof-of-concept using a DJI AGRAS T10 in a sugarcane field, it represents a baseline for offline mission planning. First, the sweep direction is optimized to minimize the total execution time of the spraying mission, directly contributing to operational efficiency. Second, the total coverage path is automatically segmented into multiple sorties according to the drone’s battery endurance. This segmentation ensures that each sortie remains within the drone’s operational limits, thereby reducing the risk of incomplete coverage due to operational endurance constraints. Furthermore, the mission’s organized subdivision enables structured, sequential execution in the Field, enhancing task management and reducing logistical complexity during deployment.

However, the approach also has some limitations. It is currently restricted to convex polygonal areas, which excludes irregular or concave-shaped fields commonly found in real-world agricultural scenarios. In addition, no static or dynamic obstacles are considered within the target area, which may limit the method’s applicability in target areas with trees, poles, or other infrastructure. Future research should extend the methodology to handle non-convex regions and integrate obstacle-avoidance mechanisms to increase robustness and practical usability. While this study demonstrated efficiency through direct metrics, future research should include sensitivity analyses to evaluate model robustness and assess how stochastic variables, such as wind fluctuations and battery discharge rates, affect mission reliability.

CONCLUSION

This research presented a novel strategy to optimize the mission planning for agricultural spraying drones, addressing operational constraints imposed by battery endurance. The proposed approach integrates coverage path planning with mission segmentation to ensure the complete spraying task is executed efficiently and within the drone’s operational endurance. A key feature of the method is optimizing the sweep direction to minimize total mission time while ensuring complete coverage of the target area. The resulting coverage path is segmented into multiple sorties, each compatible with the drone’s battery endurance. This segmentation allows for the organized execution of sub-missions, improving the manageability and operational performance of drone-based spraying tasks in the field. Beyond the methodological advances, the results provide practical guidelines for farmers and agronomists interested in adopting unmanned aerial spraying as a complementary tool in crop management.

The findings contribute to the sustainability of agricultural systems by promoting precision in input application, reducing environmental impacts from chemical overuse, and supporting the development of technologies that improve productivity while conserving natural resources. Despite the promising results, the proposed method currently presents some limitations. The approach is restricted to convex target areas and does not consider obstacles within the field. Future research will focus on extending the methodology to handle non-convex polygons and on integrating obstacle-avoidance capabilities.

ACKNOWLEDGMENTS

The author would like to thank the Universidade Federal do Triângulo Mineiro (UFTM) for the financial support provided for this research. And was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brasil - Finance code 001.

REFERENCES

  • 0
    CR-2025-0501.R2

NOMENCLATURES

  • NOMENCLATURES
    pH = Home position; h = Drone altitude; v = Drone linear velocity; δ r = Row spacing; t a = Time of battery endurance; θ = Sweep direction; m = Number of sorties; M = Target area; L j = Sweep lines; j = Coverage line number; C = Operational capacity.
  • DATA AVAILABILITY STATEMENT
    Not applicable.
  • DECLARATION OF USE OF ARTIFICIAL INTELLIGENCE
    Artificial intelligence tools were used only for language refinement of the manuscript. All AI-assisted corrections were carefully reviewed by the authors to ensure that the scientific content was not altered.

Edited by

Data availability

Not applicable.

Publication Dates

  • Publication in this collection
    14 Aug 2026
  • Date of issue
    2026

History

  • Received
    18 Sept 2025
  • Accepted
    01 Apr 2026
  • Reviewed
    08 June 2026
Creative Common - by 4.0
This is an open-access article distributed under the terms of the Creative Commons Attribution License
location_on
Universidade Federal de Santa Maria Universidade Federal de Santa Maria, Centro de Ciências Rurais , 97105-900 Santa Maria RS Brazil , Tel.: +55 55 3220-8698 , Fax: +55 55 3220-8695 - Santa Maria - RS - Brazil
E-mail: cienciarural@mail.ufsm.br
rss_feed Acompañe los números de esta revista en su lector de RSS
Notificar error