Management measures for the mitigation of spray drift of very fine droplets sprayed by a spraying robot

Abstract

Pesticide spraying is one of the most significant processes in agricultural production and one of the most complicated, risky agricultural operations. Side effects of pesticides can cause acute poisoning and serious chronic diseases in humans. Robotic spraying in agriculture is one solution to avoid human intervention. However, there has been little research on the distribution of droplets and unwanted spray drift when spraying with ground spraying robots equipped with jet spraying systems. This study analyses the downwind spray drift of three drift reduction agents (DRAs) depending on the lateral wind velocity using a ground spraying robot equipped with a jet spraying system in the field under conditionally controlled conditions. The three DRAs investigated were: DRA1 (100% anionic polymer dispersion), DRA2 (calcium dodecylbenzene sulfonate 50%, butanol 18%), and DRA3 (C10-13-alkyl derivatives, calcium salt 37%, butanol 15%). DRA solutions at a concentration of 0.1% (water as control) were sprayed with a jet spraying system and analyzed at four different droplet diameter levels ranging from VMD_preset=60 μm to 120 μm, with a change every 20 μm. The study showed that the atomization level of droplets had a significant effect on the impact of spray drift: the smaller the droplets are sprayed (VMD_preset=60–80 μm), the lower the effectiveness of DRA (spray drift can be reduced by about 2.5-fold) while spraying larger droplets (VMD_preset=100–120 μm) with DRA reduces drift by about 3.5-fold (at the lateral wind of 4 m s⁻¹). The use of DRAs also significantly impacted the reduction of spray drift. All DRA solutions were significantly more effective at low lateral winds (2–4 m s⁻¹). Moreover, the difference between the effectiveness of DRA solutions decreases with increasing lateral wind velocity from 2 to 10 m s⁻¹. In summary, the following management measures can be used to control droplet drift using a robotic jet spraying system, in order of importance: lateral wind velocity, selection of the level of droplet atomization, and the use of DRAs. This can help to find the optimal solution to ensure optimal coverage of plants with plant protection products and to minimize losses and negative environmental impacts.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-13493-3.

Introduction

By mid-November 2022, the world’s human population had reached 8.0 billion, up from an estimated 2.537 billion in 1950¹. By 2050, the number is projected to rise to 9.7 billion, peaking at nearly 10.4 billion people in the mid-2080s^1,2. For this reason, agricultural production must be increased considerably shortly to meet the food and feed demands of a rising human population and increasing livestock production. Crop protection plays an important role in ensuring crop productivity by eradicating unwanted plants (weeds), animal pests, pathogens, and viruses^3–5. It can help to grow more food on less land by protecting crops, besides raising productivity per hectare⁶. Chemical nutrient application and pesticide spraying is one of the most important agricultural production processes, but also one of the most complex, risky, and dangerous agricultural operations^7,8. The term pesticide refers to a wide range of compounds including herbicides, insecticides, fungicides, rodenticides, molluscicides, nematicides, plant growth regulators, and others⁶. According to the World Health Organization (WHO), pesticides are considered a special class of chemical compounds that are used to kill a wide range of pests including insects, rodents, fungi, and weeds^9,10. By their very nature, pesticides can be toxic to other organisms, including humans, and therefore need to be used safely and disposed of properly¹⁰.

Despite nearly 3 million tons of pesticides applied to crops annually, around 40% of the world’s food production is lost or destroyed by insects, diseases, and weeds¹¹. As mentioned above, the side effects of pesticides are of concern to humans and include acute poisonings, and severe chronic diseases^8,12. Due to these serious problems, the best solution to accomplish this task without human intervention is to use a robotic system for spraying in agriculture^8,13. Moreover, the agriculture industry is highly resource- and labor-intensive. Agricultural robots offer an opportunity to strengthen agrifood systems by addressing labor shortages and reducing CO₂ emissions¹⁴. The applications of robots are very broad − in the last few decades alone, crop production robots have been categorized into transplanting/seeding, mechanical, thermal, or chemical weed control (e.g., spraying), harvesting, pruning robots^15–17. As a result, farmers are increasingly turning to automation and robotics for spraying processes^16,18.

Usually, pesticides are sprayed with agricultural field sprayers. Based on this, spraying is carried out in several ways: with tractor-mounted boom sprayers^19,20, self-propelled field sprayers^21–23 and nowadays also by ground spraying robots^24–26 and spraying drones^27–29. The efficiency (ha h⁻¹) of boom sprayer systems mainly depends on sprayer boom width and speed of operation. For example, a boom sprayer with a 12 m wide boom and operating at 10 km h⁻¹, the efficiency would be 12 ha h⁻¹. Meanwhile, efficiency of XAG R150 robotic spraying systems and UAV XAG P100 can reach up to 5 ha h⁻¹ (at the speed of 1.2 m s⁻¹ when spray width of 12 m) and 19 ha h⁻¹ (at the speed of 13 m s⁻¹ with spray width of 3–7 m), respectively³⁰. Depending on the different applications, ground spraying robots are generally used in agricultural crop production and horticulture (mostly for nurseries, orchards, and greenhouses). The agricultural spraying robots mainly consist of chassis, spraying, and control systems³¹.

Researchers argue that all crop production processes, including spraying, are complex and varied, labor-intensive, and usually crop-specific^32–34. Fu et al.²² designed an improved self-propelled greenhouse sprayer robot that adjusts the position of the nozzle based on images and real-time remote control to avoid and control the need for different spraying locations, which reduces labor intensity and improves the automation level. A prototype system of a two-wheeled robot consisting of a mobile base, a spraying mechanism, a wireless controller to control the movement of the robot, and a camera to monitor crop health and growth and to detect pests in agriculture was developed by Ghafar et al.¹⁸. The spraying system was designed to provide only the optimal required quantity of fertilizers and pesticides to individual plants, thus reducing the wastage of liquid fertilizers or pesticides.

One of the most important elements of both conventional and robotic sprayers is the spraying system with different types of nozzles. Full cone nozzles have been investigated in precision variable-rate spraying robots using single 3D LIDAR in orchards³⁵, while electromagnetic nozzles have been investigated in a high-precision patch-spraying robotic system based on weed maps³³. An evaluation of the spray generated by a greenhouse spraying robot containing a full cone nozzle and droplet size distribution affecting drift was carried out and discussed by Kalantari et al.³⁶. Rincón et al.³⁷ optimized the spray profile of a remotely controlled spraying robot and an aerial assist system in the laboratory. In this research spray performance using two different nozzle sets (full cone and hollow cone) with and without air assist was compared with the performance of a sprayer in a greenhouse tomato crop. Most of the developed robots can use standard flat nozzles and air-injector nozzles²³, air supply spraying systems^21,37 or jet spraying systems with air-assist atomizing nozzle^30,38,39.

In addition to selecting a suitable spraying system, one of the main problems in the spraying industry has always been spray drift. It is defined as the movement of pesticide droplets or solid particles outside the target area being treated due to lateral wind conditions^11,40. It is impossible to avoid spray drift completely, but it can be minimized by using best management practices. The main factors influencing such spray drift are lateral wind velocity and direction, air temperature, relative air humidity, nozzle type, spray droplet size, spray angle, spraying height above the crop, travel speed, and formulation of spray adjuvants^40–44. Many measures have been developed to reduce the effects of pesticides due to spray drift, which can be divided into technical, technological, organizational, and chemical measures⁴⁵. Chemical measures, more specifically drift reduction agents (DRAs), defined as the material used in liquid spray mixtures that reduce drifting fine particles, are currently being studied quite extensively⁴⁶. DRAs can be categorized into 3 main groups according to their constituent substances: solvents, synergists or inert ingredients, and surfactants^47,48. Surfactants are divided into anionic, cationic, non-ionic, and amphoteric⁴⁹. Anionic surfactants usually consist of hydrophilic functional groups such as phosphate, sulfate, or carboxylate groups with a sodium or calcium counterion. The most commonly used compound is linear alkyl benzene sulfonate⁵⁰. Cationic surfactants such as amines or cetrimonium bromide (CTAB) have cationic functional groups, often quaternary ammonium ions, which belong to the hydrophilic functional group⁴⁹. Non-ionic surfactants do not ionize in aqueous solutions because their hydrophilic groups are of a non-separating type, either alcohol or glycol and are uncharged and hydrophilic⁵¹. Amphoteric surfactants can have both positive and negative charges and are very similar in functionality to non-ionic surfactants and their dissociation is strongly influenced by the pH of the medium⁵².

The main types of DRAs used in agricultural practice are non-ionic surfactants, organic silicates, mineral sources, and plant source synergists⁵³. A detailed study of seven DRA properties that help to understand the process of drift formation in agricultural spraying was conducted by Jomantas et al.⁴⁵. Two DRA solutions based on calcium dodecylbenzenesulfonate and benzenesulfonic acid, C10-13-alkyl derivatives, and calcium salt, were developed and shown to be the most successful drift reduction agents^45,54. Drift measurements were performed using the open circuit-type wind tunnel and in the field under conditionally controlled conditions. To substantiate the efficacy of these DRAs, it is appropriate to carry out an in-depth study on their spraying with a ground spraying robot.

While all the above studies on robotic spraying aim to substantiate the spraying performance, detailed studies on spraying drift using jet spraying systems with air assistance atomizing nozzles are still lacking. Despite the considerable advantages of robotic spraying, they are likely to experience unwanted spray drift, just like conventional spraying systems. This is particularly the case if the robotic spraying system uses fine spraying, i.e. the use of a jet spraying system with air assistance atomizing nozzle.

The aim of this study was to investigate the effectiveness of management measures for the mitigation of spray drift, specifically by comparing the downwind spray drift of three drift reduction agents and evaluating their performance across varying lateral wind velocities, when applied by a spraying robot jet spraying system in the field under conditionally controlled conditions.

Materials and methods

Robotic sprayer

Experimental studies on the robotic sprayer were carried out between 2022 and 2023 at Vytautas Magnus University Agriculture Academy, Department of Agricultural Engineering and Safety (Lithuania). The XAG R150 spraying robot (XAG, China) (Fig. 1) consists of the following main elements: a liquid tank (1), a control unit (4), a frame (5), two electric motors, two peristaltic liquid pumps, and two XAG JetSprayer™ systems (2) (jet spraying systems). The XAG R150 is equipped with the SUPERX3 Pro RTK control system, which is very precise and can achieve an accuracy of 1 cm. The robot can operate in several different modes: point mode, i.e. following a route, “Follow me” mode, i.e. following the person who controls it, and remote-control mode. The robot’s parameters can be controlled and changed using an app on the smartphone and the control panel. The robotic sprayer can travel at three different velocities: 0.4 m s⁻¹, 0.8 m s⁻¹, and 1.2 m s⁻¹. The robot has a minimum turning radius of 0.7 m. It is driven by 4 wheels, which are chain-driven by 2 brushless electric motors transmitting a maximum torque of 1000 Nm. Power is provided by two B13860S smart batteries with a capacity of 18Ah (865.8 Wh), a voltage of 48.1 V, and a runtime of 4 h. Charging time is 15–18 min.

Fig. 1.

Schematic view of robotic sprayer: 1 – liquid tank; 2 – Smart Pan Tilt with XAG JetSprayer™ system (jet spraying system); 3 – drive wheels; 4 – control unit, 5 – frame; α₁ – vertical angle of movement of the jet spraying system; α₂ – horizontal angle of movement of the jet spraying system.

The nozzles of the robotic jet spraying system can change position in two planes along different trajectories. In this spraying system, both nozzles can move in a 200° vertical trajectory and 290° horizontal trajectory. The maximum spray width can reach 12 m, and the diameter of the spray droplets can range from 60 μm to 200 μm. Each nozzle is supplied by a separate peristaltic pump. The pump has a capacity of 2.4 l min⁻¹. The liquid tank capacity is 100 l.

Jet spraying system and droplet atomization in it

The jet spraying system consists of the following main components (Fig. 2): two axial fans with impellers (3 and 5) driven by two separate electric engines (4 and 6), a liquid inlet channel (11), and a nozzle (10). The whole jet spraying system is connected to the Smart Pan Tilt spraying robot via an attachment (1).

Fig. 2.

Jet spraying system components: 1 – attachment to the Smart Pan Tilt; 2 – back hood with sifter; 3 and 5 – impellers; 4 and 6 – electric engines, 7 – front cover; 8 – nozzle holder; 9 – nozzle cap; 10 – nozzle; 11 – liquid inlet channel.

An electric engine (6) drives an impeller (5) with a diameter of 70 mm, which generates the initial airflow and delivers it to the conical nozzle holder (8) and finally to the nozzle itself (10). After entering the nozzle holder (8) at a high velocity (max. 160 m s⁻¹), the supplied air bounces against the nozzle retaining cap (9) and starts to move through the channels in the nozzle structure. At the same time, the liquid is fed from the peristaltic pump through the channel (11) into the nozzle (10). This airflow, indicated by the dark green arrows, is necessary for the atomization of the spray solution into droplets. Thus, before the liquid to be sprayed enters the environment, it is mixed with a high-velocity air stream. By varying the rotation velocity of the electric engine and the fan impeller, the air velocity is also varied using a control unit (4) (Fig. 3). The higher the air velocity, the more intense the droplet atomization, resulting in a smaller VMD of the sprayed droplets.

Fig. 3.

Jet Sprayer working scheme: 1 – nozzle cap; 2 – nozzle holder; 3 – liquid inlet channel; 4 – electric engine with control unit; 5 – impeller, 6 – electric engine, 7 – Jet Sprayer front cover; 8 – nozzle.

An electric engine (4) at the rear of the jet spraying system, which rotates a 150 mm diameter impeller (3), provides another secondary airflow that blows the droplets sprayed from the nozzle cavity (10). This airflow moves between the components of the spraying system and the inner part of the Jet Spraying System’s front cover (7). This airflow can be used to modify the flight path of the droplets and the length and width of the target area.

This spraying system offers 8 atomization levels (Table 1), which can be set by varying the rotational velocity of the fan impeller (5) (Fig. 3). The atomization level of the droplets also changes the VMD_preset (volume median diameter), which is a measure of the size dispersion of the sprayed droplets.

Table 1.

Droplet atomization levels.

Droplet atomization level	Impeller rotational velocity, natom, RPM	Air velocity, vatom, m s−1	VMDpreset, µm	Note
1	10,625	20	200	Atomization of droplets with the lowest intensity
2	21,250	40	180	-
3	31,875	60	160
4	42,500	80	140
5	53,125	100	120
6	63,750	120	100
7	74,375	140	80
8	85,000	160	60	Atomization of droplets with the highest intensity

The detailed nozzle design and the directions of liquid and air flow are shown in Fig. 4. The airflow (indicated by the green arrows) rises upwards through the cavities at the bottom of the nozzle (2) and bounces against the nozzle cap (1) at the top. The airflow (3) travels through the cavities in the nozzle structure and reaches the spray liquid opening. The liquid flow is mixed with the moving airflow and the droplets are broken up (the spray of broken droplets is shown by the blue arrows).

Fig. 4.

Jet Sprayer droplet atomization: 1 – nozzle cap; 2 – nozzle; 3 – airflow movement direction; 4 – nozzle hole (liquid channel).

Field experiments on the effect of environmental factors (lateral winds) on droplet drift

The field experiments were carried out between 10 am and 5 pm. The air temperature was 19 ± 3 °C and the relative humidity was 70 ± 5%. A stand was constructed for the tests to create an artificial airflow – a lateral wind (Figs. 5 and 6). The stand consists of two axial fans ML 1004 DT (Electrovent, Italy), impellers (diameter 1000 mm) consisting of 10 plastic blades, electric engines 7SM3 160L4 (power, 15 kW, and rotation, 1465 RPM) (Smem, Monza, Italy), and an airflow straightener. The rotation velocity of the fans was varied by Delta VFD-C2000 voltage frequency converters. They were used to vary the airflow (crosswind) velocity from 2 m s⁻¹ to 10 m s⁻¹ at 2 m s⁻¹ intervals. Lateral wind velocity measurements were also taken⁴⁵.

Fig. 5.

Setup for field spray drift experiments with a robotic sprayer: 1 – fans and airflow straighteners, 0.25 m in length; 2 – robotic sprayer; 3 – airborne spray drift setting rod with water-sensitive paper attachment points; 4 – meteorological station; 5 – ground spray drift stands with water-sensitive paper attachment points; 6 – frequency converters; α – high of ground drift stands.

Fig. 6.

The real view of field spray drift experiments with a robotic sprayer.

The drift of downwind droplets was studied in the horizontal plane. For the horizontal droplet coverage studies (Fig. 5), stands (5) were used. They were positioned at 1.0, 2.0, 3.0, 4.0, 5.0, 7.5, 10.0, 15.0, and 20.0 m away from the target area of the robotic sprayer (2) (according to ISO 22866:2005). The height of the stands from the soil surface was 0.1 m. Water-sensitive papers were attached to the top of them. A meteorological station was placed 5.0 m from the target area to record the environmental conditions.

During the spraying experiments, the robot’s travel speed was chosen to be 0.4 m s⁻¹. This driving speed was chosen because the robot will spray a greater amount of liquid through the lateral wind zone created by the fan. The spray rate was set at 4.8 l ha⁻¹. The spraying parameters were adjusted so that the width of the target area of the robot was 2.8 m. During the experiments, the robot’s spray components moved vertically from − 1° to 29° and horizontally from 85° to 120° (Fig. 1). The spray rate chosen was 4800 ml min⁻¹.

The robotic spraying experiments were carried out at 4 different atomization levels determined before the study. These are figured out by varying the rotational speed of the fan impeller (5) (Fig. 2) from VMD = 60 μm (at the highest atomization level, air velocity is v_atom=160 m s⁻¹ and impeller rotational speed is n_atom=85000 RPM) to VMD = 120 μm (at the lowest atomization level, v_atom=100 m s⁻¹, n_atom=53125 RPM), increasing the VMD of the sprayed droplets by increments of 20 μm. The maximum secondary air velocity (level 8, v_booster=16 m s⁻¹, n_booster=9000 RPM) was also determined, which is induced by a fan (3) to control the target area and the droplet trajectory (Fig. 2).

Drift reduction agents (DRAs)

Three DRAs (0.1% concentration) and water (control) with different chemical compositions were used in the field. Their effect on reducing the influence of lateral wind on the movement of sprayed droplets was investigated. The properties of DRA1, DRA2, and DRA3 and water were also tested and determined in the laboratory. The static surface tension was measured with a Digital Tensiometer Easy Dyne (Krüss, Germany) using the Wilhelmy plate method. The tension determined was equal to the average of three measurements with a deviation of less than 1%. The dynamic surface tension was measured using a bubble pressure tensiometer BP50 (Krüss, Germany). At least three replicate measurements were made for each liquid tested. The temperature in the laboratory at the time of the measurements was 22 °C. The viscosity of the solutions was determined (at 20 °C) using an SVM™ 3000 Stabinger Viscometer™ from Anton Paar. DRA densities were determined by weighing a 100 ml sample.

Anionic DRA1 (100% anionic polymer dispersion) with a density at 100% solution of 1.03 g cm⁻³, a viscosity of 2000 mPa s, a static surface tension of 31.6 mN m⁻¹, and a dynamic surface tension of 69.7 mN m⁻¹ was measured over a 50 ms interval and 69.2 mN m⁻¹ measured over a 100 ms interval.

Anionic DRA2 (calcium dodecylbenzene sulfonate 50%, butanol 18%) with a density at 100% solution of 1.10 g cm⁻³, a viscosity of 2300 mPa s, a static surface tension of 30.5 mN m⁻¹ and a dynamic surface tension was measured over a 50 ms interval of 64.6 mN m⁻¹ and a 100 ms interval of 63.7 mN m⁻¹.

Non-ionic DRA3 (C10-13-alkyl derivatives, calcium salt 37%, butanol 15%) with a density at 100% dilution of 1.03 g cm⁻³, a viscosity of 700 mPa s, static surface tension of 32.4 mN m⁻¹ and a dynamic surface tension was measured over a 50 ms time interval of 69.6 mN m⁻¹ and a dynamic surface tension measured over a 100 ms time interval of 69.5 mN m⁻¹.

The properties of water were also determined. The density was 1.10 g cm⁻³, the viscosity was 1.0 mPa s, the static surface tension was 72.0 mN m⁻¹ and the dynamic surface tension was 71.6 mN m⁻¹ measured over 50 ms, and 71.5 mN m⁻¹ measured over 100 ms.

Droplet analysis of sprayed liquid

After the experiments, water-sensitive papers (Syngenta, water-sensitive paper 26 × 76 mm, Switzerland) are collected and allowed to dry. The papers are then scanned with a scanner and the image is converted to a 600-dpi monochrome image and processed with the computer program DepositScan⁵⁵. Using the program, an area of 1 cm² is selected, and droplet coverage (%) is obtained.

Statistical analysis

Data points are presented as mean values with confidence levels calculated at 95% probability with the statistical software Statistica 10.0.

Results

The spraying experiments investigated the drift of spray droplets in a 20 m-long study area outside the target area boundary under different lateral velocities. The experiments were started at an airflow rate of 2 m s⁻¹ and increased by 2 m s⁻¹ intervals until an airflow rate of 10 m s⁻¹ was reached. Most of the literature and the requirements of the spraying operation mention that spraying operations are not recommended at wind velocities above 4 m s⁻¹. However, it is not uncommon to encounter wind gusts above the recommended value during a spraying operation. Therefore, spraying experiments were also carried out at lateral velocities of 6–10 m s⁻¹. Three different droplet reduction chemicals, DRAs, were tested. To determine their effectiveness and the effect of different droplet diameters on VMD_preset drift, droplets with diameters ranging from 60 to 120 μm (in increments of 20 μm) were set up in the robot system.

At the lowest lateral velocity (2 m s⁻¹), it was observed that droplets at different VMD_preset are carried at different distances (Fig. 7). At VMD_preset=60 μm, it was observed that in all cases studied (all DRAs and water), the droplets of the sprayed liquid drifted to a collector located at 7.5 m from the spraying area. At the same wind velocity, the droplet drift was reduced with DRA compared to water. At all distances, the difference in performance varied, e.g. at L = 3 m from the spraying area, the coverage of the droplets by the sprayed water was 4.6%, 3.6% for DRA1 (1.3-fold reduction in drift), 2.7% for DRA2 (1.7-fold reduction in drift), 2.5% for DRA3 (1.8-fold reduction in drift). Comparing the DRAs with each other, the difference between DRA2 and DRA3 is very small, only 0.2% points, but the difference is more pronounced when compared to DRA1. When spraying the solution and applying DRA1, about 25% more droplets were deposited compared to DRA2 and about 30.6% more droplets compared to DRA3.

Fig. 7.

The influence of DRA on the drift of downwind droplets outside of the target area at a lateral wind velocity of 2 m s⁻¹ when a droplet atomization level set in the spraying system guarantees VMD_preset=60 μm. Data are presented as mean value ± confidence level (95%).

If, under the same conditions, the diameter of the droplets to be sprayed is increased in VMD_preset, the droplet drift is reduced. Increasing the droplet diameter to VMD_preset=120 μm resulted in all DRA and water droplets being drifted to a collector located 4 m from the spraying area (Fig. 8). As in the previous case, there are visible differences in the effectiveness of the formulations compared to water, but in all cases the DRAs used are effective. At 3 m from the spraying area, the coverage of the droplets by the water spray was 0.4%, while for all DRAs it was 0.1% (drift reduced 4-fold). There was no difference between the results of the DRAs at 3 m from the spraying area.

Fig. 8.

The influence of DRA on the drift of downwind droplets outside of the target area at a lateral wind velocity of 2 m s⁻¹ when a droplet atomization level set in the spraying system guarantees VMD_preset=120 μm. Data are presented as mean value ± confidence level (95%).

A comparison between sprayed solutions and water, under the same conditions but with a change in VMD_preset, showed that increasing the droplet diameter from 60 μm to 120 μm resulted in 11.5-fold less water entrainment – 36-fold less for DRA1, 27-fold less for DRA2 and 25-fold less for DRA3. This means that at low lateral velocities (2 m s⁻¹), increasing the droplet diameter results in a significant reduction in spray drift. There were also analogue studies carried out with VMD_preset at 80 μm and 100 μm and the results are referred to in the Supplementary material (Figs. S1−S2).

At a lateral velocity of 4 m s⁻¹ and with the finest droplet separation (VMD_preset=60 μm) detected by the robot jet spraying system, it was found that the drift of the sprayed droplets using the tested DRAs was recorded within 10 m of the spraying area (Fig. 9). In the control case (water-only spraying), droplet drift was up to 15 m. It was observed that all three DRAs reduced droplet drift. As the distribution of droplets varied between the different locations in the field, a collector at L = 4 m from the target area was chosen to analyze the effect of the DRAs on the downwind drift of sprayed droplets. The water coverage of the paper on the collector was 8.0% – 5.5% for DRA1 (1.5-fold reduction in drift), 2.2% for DRA2 (3.6-fold drift reduction), and 2.8% for DRA3 (2.9-fold drift reduction). The results show that the droplets sprayed with DRA1 have the highest drift of all the DRAs tested. DRA2 has 2.5-fold less droplet drift (or 60%) than DRA1, while DRA3 has about 1.9-fold less drift (49.1%) compared to DRA1. There was no significant difference between DRA2 and DRA3 as they differed by only 0.6% units (points).

Fig. 9.

The influence of DRA on the drift of downwind droplets outside of the target area at a lateral wind velocity of 4 m s⁻¹ when a droplet atomization level set in the spraying system guarantees VMD_preset=60 μm. Data are presented as mean value ± confidence level (95%).

Increasing the VMD_preset to 120 μm and the lateral velocity to 4 m s⁻¹, it was observed that for both the control and the DRAs, the sprayed droplets drifted 7.5 m from the spraying area (Fig. 10). For the analysis of the effect of DRAs on the reduction of droplet drift, a collector at L = 4 m from the target area was chosen. The coverage was 1.7% for the control, 0.9% for DRA1 (drift reduced by 1.9-fold), 0.5% for DRA2 (drift reduced by 3.4-fold), and 0.3% for DRA3 (drift reduced by 5.7-fold). Comparing the DRAs, it can be observed that in the spraying experiments, a higher proportion of the solution droplets sprayed by DRA1 was carried downwind than by DRA2 and DRA3. Approximately 1.8-fold more (or 37.5%) droplets were carried away compared to DRA2 and 3-fold less (or 66.7%) compared to DRA3. A comparison of DRA2 and DRA3 showed a 1.7-fold (40%) reduction in drift in DRA3.

Fig. 10.

The influence of DRA on the drift of downwind droplets outside of the target area at a lateral wind velocity of 4 m s⁻¹ when a droplet atomization level set in the spraying system guarantees VMD_preset=120 μm. Data are presented as mean value ± confidence level (95%).

Increasing the VMD_preset of the sprayed droplets from 60 μm to 120 μm shows a reduction of about 4.7 times (78.8%) for the control. For DRA1, the droplet drift was reduced by 6.1-fold (83.6%), for DRA2 it was reduced by 4.4-fold (77.3%), and for DRA3 drift was reduced by 9.3-fold (89.3%). Similar tests were carried out for VMD_preset at 80 μm and 100 μm and the results are given in the Supplementary material (Figs. S3−S4).

A lateral wind of 10 m s⁻¹ created by the airflow generation stand showed that droplets sprayed with water and solutions containing DRAs drifted 20 m outside the target area (Fig. 11). After the experiments, it could be observed that the dispersion and differences between the sprayed solutions and the control in the study field were not uniform. To investigate the results of the spray experiments and to determine the influence of the DRAs on the drift of the sprayed droplets, a collector located L = 2 m from the target area was selected. For the control case, the VMD_preset=60 μm resulted in a 35.6% drift, for the solution sprayed with DRA1 the drift was 30.3%, with a 1.2-fold (or 14.9%) reduction in the drift of the sprayed droplets, for DRA2 the drift decreased by 1.3-fold (or 21.1%) for DRA2 28.1%, and for DRA3 the drift decreased by 1.5-fold (or 34.8%). Looking for differences between the DRAs, it was found that the droplet drift of the sprayed DRA2 solution is about 1.09-fold (about 7.3%) lower than that of DRA1 and that of the DRA3 solution is about 1.3-fold (22.8%) lower than that of DRA1. A comparison of the results obtained with DRA2 for DRA3 showed that DRA contributes to a reduction of about 1.2-fold (16.7%) in the drift of sprayed droplets.

Fig. 11.

The influence of DRA on the drift of downwind droplets outside of the target area at a lateral wind velocity of 10 m s⁻¹ when a droplet atomization level set in the spraying system guarantees VMD_preset=60 μm. Data are presented as mean value ± confidence level (95%).

When the target area was exposed to a lateral wind of 10 m s⁻¹, but the VMD_preset of the sprayed droplets was increased to 120 μm, it was found that the droplets sprayed by the control and DRA solutions drifted to 20 m outside from the target area. To look for differences in the drift of sprayed droplets, a collector located 2 m from the target area was chosen (Fig. 12). The droplet coverage was 21.8% for the control, 20.0% for DRA1 with a decrease of 1.09-fold (or 8.3%), 18.5% for DRA2 with a drift reduced by approximately 1.18-fold (or 15.1%), and 16.1% for DRA3, with a decrease of about 1.35-fold (or 26.1%) compared to the control. When analyzing the difference between the DRAs, DRA1 has the highest droplet drift. The comparison shows that DRA2 has about 1.08-fold (7.5%) fewer droplet droplets than DRA1 and DRA3 – about 1.2-fold (19.5%) fewer than DRA1. The results show that DRA3 has the lowest downwind drift. Compared to the results obtained for DRA2, DRA3 has about 1.2-fold (13.0%) fewer droplets carried downwind.

Fig. 12.

The influence of DRA on the drift of downwind droplets outside of the target area at a lateral wind velocity of 10 m s⁻¹ when a droplet atomization level set in the spraying system guarantees VMD_preset=120 μm. Data are presented as mean value ± confidence level (95%).

A comparison was made to see the difference when changing the VMD_preset value from 60 μm to 120 μm and at 10 m s⁻¹ lateral velocity. For the control, increasing the VMD_preset value of the sprayed droplets resulted in a 1.6-fold (38.8%) reduction of droplet drift, for DRA1 the drift was reduced by 1.5-fold (about 34%), for DRA2 drift it was reduced by 1.5-fold (about 34.2%) and for DRA3 the drift was reduced by 1.4-fold (31.2%). The remaining experiments were performed both at VMD_preset at 80 μm and 100 μm and lateral velocities of 6 m s⁻¹ and 8 m s⁻¹, and the results are given in the Supplementary material (Figs. S5−S14).

To assess the effectiveness of the three DRAs as a function of the lateral velocity (varying from 2 m s⁻¹ to 10 m s⁻¹), the coverage of a collector 4 m from the target area with droplets of sprayed liquid was analyzed. At a VMD_preset of 120 μm and a lateral wind velocity of 2 m s⁻¹, the droplet coverage on the water-sensitive paper was consistently 0.2% across all tested solutions (water and DRAs). In contrast, for all DRAs used, the coverage did not exceed 0.1% (Fig. 13). The difference between DRA and water becomes more pronounced as the lateral velocity increases. Increasing the lateral velocity to 10 m s⁻¹, the water droplet coverage of the paper on the collector was 14.2%, 12.2% for DRA1, 12.9% for DRA2, and 12.6% for DRA3. The water droplet coverage of the paper was about 11% higher than that of any DRA.

Fig. 13.

The influence of the lateral wind velocity and DRA on the drift of downwind droplets 4 m outside of the target area when a droplet atomization level set in the spraying system guarantees VMD_preset=120 μm.

At the lower droplet atomization level (VMD_preset=100 μm) set on the robot, at a lateral velocity of 2 m s⁻¹, the water droplet coverage of the water-sensitive paper on the collector was 0.4%, 0.2% for DRA1 and 0.1% for DRA2 and DRA3 (Fig. 14). The dependence of the variation of droplet deposition on the influence of lateral winds can be approximated by power functions (Fig. 14). The difference between the three DRAs is insignificant, while the difference between water and DRA is relatively pronounced. However, the difference is significantly reduced at a lateral velocity of 10 m s⁻¹. In comparison, at 8 m s⁻¹ lateral velocity, DRA has a lower drift than water by about 39%, while at 10 m s⁻¹, it is only about 11% lower.

Fig. 14.

The influence of the lateral wind velocity and DRA on the drift of downwind droplets 4 m outside of the target area when a droplet atomization level set in the spraying system guarantees VMD_preset=100 μm.

When the robotic jet spraying system was used to detect even more intense droplet atomization (VMD_preset=80 μm) and the results were analyzed at L = 4 m from the spray area and a lateral velocity of 2 m s⁻¹, the water coverage of the water-sensitive paper was 2.7% for the robotic spraying system, 0.4% for DRA1, 0.1% for DRA2 and 0.1% for DRA3 (Fig. 15). In this case, the water drift compared to DRA is significantly increased and is higher (27.7-fold higher than in DRA) than in the previous cases studied (3.4-fold higher at VMD_preset=120 μm and 4.5-fold higher than in DRA at VMD_preset=100 μm). It can be noted that the difference between water and DRA is quite significant with increasing wind velocity and that at a lateral wind velocity of 10 m s⁻¹, the water coverage of the water-sensitive paper was already 24.3%, while it was 16.6%, 16.3%, and 15.6% for DRA1, DRA2, and DRA3, respectively. Taking all DRAs together, it can be said that 34% less of them were taken away compared to water.

Fig. 15.

The influence of the lateral wind velocity and DRA on the drift of downwind droplets 4 m outside of the target area when a droplet atomization level set in the spraying system guarantees VMD_preset=80 μm.

After performing experiments with the smallest droplets (60 μm) of the jet spraying system of the robotic sprayer, and the spraying process under the influence of a lateral wind of 2 m s⁻¹ velocity, the water coverage L = 4 m from the target area reached 3.2%, while for DRA1–2.6%, DRA2–1.7%, DRA3–2.3% (Fig. 16).

Fig. 16.

The influence of the lateral wind velocity and DRA on the drift of downwind droplets 4 m outside of the target area when a droplet atomization level set in the spraying system guarantees VMD_preset=60 μm.

The water drift compared to the DRA increased less compared to the previous case. In addition, the difference between the sprayed water droplets and the DRA is not so big (the water drift is 1.5-fold higher than the DRA). As can also be seen in Fig. 16, with increasing wind velocity, the difference between water and DRA also increased. For comparison, the results at a lateral velocity of 6 m s⁻¹ show that the water droplet coverage reached 10.7%, while DRA1–7.4%, DRA2–8.2%, and DRA3–8.5%. Thus, about 1.3-fold more water was drifted. An analogous comparison with a lateral velocity of 10 m s⁻¹ resulted in water coverage of 24.8%, DRA1–18.8%, DRA2–17.0%, and DRA3–16.9%. Water drifted 1.4-fold more than DRAs. According to the obtained results, it can be stated that the influence of DRA on the drift of very fine droplets is only slightly reduced at higher lateral wind velocities.

Discussion

The existing research on robotic ground spraying focuses on single factors, such as real-time pest detection systems^18,56, real-time disease identification^24,57, robotic sprayer nozzle positioning²², and optimal pesticide application rates¹⁸ without taking into account the effects of the interaction of other influential factors. Nevertheless, the effects of different adjuvants used in spray studies on drift when spraying with different nozzles in a wind tunnel^58,59 or with agricultural drones⁶⁰ are numerous. This limitation called for the need to conduct experimental research that showed spray droplet deposition characteristics with lateral wind and droplet size characteristics under conditionally controlled conditions.

According to Weicai and Panyang⁶¹, when conducting drift experiments in real field conditions, it is quite difficult to obtain adequate results because of the constantly changing meteorological factors such as wind velocity, direction, ambient temperature, and humidity. For this reason, in a previous study⁴⁵, a wind generator test stand under conditionally controlled conditions was developed to simulate external spray conditions by controlling the velocity and direction of the airflow (wind). This was an important tool to investigate the drift characteristics of spray fluids and avoid many of the shortcomings of field experiments. This test stand was also used in this study, where the lateral velocity was varied over a wide range from 2 to 10 m s⁻¹ using ground spraying robots with a jet spraying system. It should be noted that research on lateral wind velocity and various drift reduction agents’ impact on spray drift of very fine droplets with these robotic systems is still lacking in the scientific literature.

Yang et al.⁶² have analyzed the influence of lateral wind and electrostatic voltage on the spray drift of an electrostatic sprayer. The study was conducted with a self-propelled orchard sprayer with a spray boom. By increasing the lateral wind velocity (from 1 to 8 m s⁻¹) and electrostatic voltage, the spray drift increased similarly to the present study. This was also evident from the results of Sun et al.⁶³ for the spraying of plant protection products, focusing on citrus fruits, where the effect of different lateral velocity (from 0 to 3 m s⁻¹) and nozzle inclination angle (from 0 to 45°) on droplet penetration was investigated. Previous agricultural spraying studies have also confirmed that spray drift is very sensitive to the wind velocity during the aerial spraying process⁶⁴ or in a wind tunnel⁴⁰: a small variation in wind velocity leads to a relatively large difference in drift distance.

The adjuvants DRA1, DRA2, and DRA3 investigated in this study belong to the largest group of adjuvant-surfactants. Our results demonstrate the importance of DRA addition at the concentration of 0.1% and atomization jet spraying system in the analysis domain. DRA1 and DRA2 are classified as ionic surfactants and DRA3 – as non-ionic. The term surfactant is a short form for surface active agent derived from the ability of the material to accumulate at the interface between the polar (water) and non-polar (air) phases – the surface layer – and to reduce surface tension forces. Surfactants are also used to increase or decrease the solubility of one fluid in another. Reducing solubility often results in more stable fluid boundaries⁶⁵. Previously, several scientists who have investigated the role of adjuvants have reported positive effects of adjuvants on drift control⁶⁶. Moreover, droplet size is also one of the most important indicators for classifying sprays. American Society of Agricultural and Biological Engineers (ASABE) has updated and adopted the British Crop Protection Council (BCPC) classification standard and has classified droplets into extremely fine, very fine, fine, medium, coarse, very coarse, extremely coarse, and ultra-coarse⁶⁷. The amount of drift loss is related to the proportion of very fine and fine droplets⁶⁸. The use of a robotic jet spraying system during the atomization process produces fine and very fine droplets with the highest drift, but low droplet bounce, so it is best to apply such droplets to the crop^42,69. The results reveal that the drift of droplets sprayed by the ground spraying robot depends on lateral wind velocity and droplet atomization, which determines the size of droplets in the robotic spraying system.

The nozzle is a very important part of the pesticide application process. The liquid sheet produces ruptures through various mechanisms and forms droplets in a process known as atomization. Atomization, referring to the conversion of bulk liquid into a collection of drops (i.e. a spray), often occurs after the liquid passes through a nozzle. Numerous devices have been developed to generate spray flows and they are generally designated as atomizers or nozzles. Although atomization does not normally indicate that the liquid particles are reduced to atomic sizes, the spray drops from atomization can be very small⁷⁰. This study also found that droplet drift depends on the droplet breakup in the robotic spraying system (atomization), which determines droplet size.

From a practical point of view, droplet diameter is strongly affected by nozzle type and operating pressure. All technologies that generate larger droplets will benefit from drift reduction: low pressure, pre-orifice, deflector, induction chamber, and air inclusion⁴⁰. The main function of the nozzle is to atomize the pesticide into droplets and, under the action of external forces, deliver the liquid to the target crops. Droplet size and velocity directly influence droplet movement. It also affects the drift and deposition rate, further influencing pesticide application efficiency and pest control effectiveness. Small droplets have favorable deposition and coverage effects but are susceptible to evaporation and drift, which may influence their application efficiency⁷¹. It is suggested that the droplet size in agricultural sprays should be between 100 and 400 μm to minimize the drift and maintain a good deposition and coverage⁴². It is also noted that spray can produce droplets smaller than 150 μm in diameter, reduce spray efficiency, and potentially affect surrounding non-target organisms as fine spray can drift away due to ambient wind⁶⁶. The droplet diameter distribution as well as the spatial drift loss are jointly influenced by the nozzle structure, the operating pressure, and the pesticide liquid⁶⁸.

In this study, it was found that at lateral air velocities of 2 m s⁻¹, larger DRA droplets (with VMDs of 100–120 μm) were carried up to 4 m from the spray area (Fig. 8). Small droplets (60–80 μm VMD) drifted up to 7.5 m from the spray area (Fig. 7). This highlights the critical role of droplet size in mitigating drift for ground-based robotic applications. In comparison, studies on unmanned aerial vehicles (UAV), such as Wang et al.⁶⁰ and Semenišin et al.⁷², emphasized the significance of VMD and DRA effectiveness in managing spray drift. However, their findings, derived from applications at greater flight heights, highlight a different set of drift dynamics influenced by greater dispersal height, prop wash, and complex ground-induced vortexes. For example, Semenišin et al.⁷² found that while DRAs could reduce UAV drift by up to 58.4% in some cases and improve coverage, operational altitudes of 1.5 m provided minimal improvements due to susceptibility to ground-induced vortex-aided droplet dispersion, suggesting that maintaining close proximity to the ground is not recommended for emerging crops with UAVs. This contrasts with our ground robot’s ability to maintain direct, very low-level application. The direct proximity of ground spraying robots to the target canopy, as used in our experiments, inherently offers greater control over droplet placement and potentially reduces the initial dispersal height that can exacerbate drift in aerial applications. While UAVs benefit from avoiding soil compaction and can quickly cover challenging terrain, their higher application altitude (e.g., 1.5–2.0 m in Wang et al.⁶⁰, and similarly noted by Semenišin et al.⁷² often necessitates a greater reliance on very fine droplets for coverage and can be more susceptible to drift from horizontal wind shear over larger distances. Kalantari et al.³⁶ further underscore the importance of drift reduction strategies for aerial platforms, arguing that drift can be significantly reduced by changing nozzle type and adding pesticide adjuvant to the spray solution. Their research on liquid sheet breakup characteristics with different DRAs and commercial spray nozzles (XR, AIXR, and TXVK) for drones, finding significant effects of nozzle type and solution on breakup, reinforces the universal importance of formulation and nozzle selection across both ground and aerial spraying technologies for effective drift management. Our study, focusing on a jet spraying system, complements this by showing how a ground robot can leverage precise VMD control and DRAs for localized drift mitigation. Ultimately, while both robotic ground and aerial systems contribute to precision agriculture, the ground-based approach demonstrates a strong capability for minimizing off-target drift due to its lower application height and more controlled interaction with the crop environment, particularly when larger droplets and DRAs are employed.

However, there are not many comprehensive studies on the use of ground-spraying robots. In particular, there is a lack of research on the drift of such robots, as well as on the use of different anti-drift agents.

Conclusions

1. Studies have shown that lateral wind velocity v has the greatest effect on droplet drift of all the parameters investigated in this study. By increasing the lateral wind velocity from 2 m s⁻¹ to 10 m s⁻¹, the spray drift progressively increases. A particularly rapid increase in spray drift was observed from v = 6 m s⁻¹. At a wind velocity of 2 m s⁻¹ (sprayed droplets volume median diameter VMD_preset=60 μm), at 4 m from the target area, the coverage of water-sensitive paper with water was about 3.2%, 8% at v = 4 m s⁻¹, 8.3% at 6 m s⁻¹, 20.7% at 8 m s⁻¹, and 24.8% at 10 m s⁻¹. By increasing the VMD_preset to 120 μm, the coverage decreased to 0.2%, 1.7%, 3.9%, 7.4%, and 14.2%, respectively.

2. It has been found that the spray drift can be controlled by modifying the properties (surface tension and viscosity) of the spray solution, i.e. by using DRAs (chemical drift reduction additives) in the spray solution. Regardless of wind velocity and droplet atomization level (VMD_preset), the spraying of DRAs resulted in a lower volume of spray liquid being drifted. At a lateral wind velocity of v = 10 m s⁻¹ (VMD_preset=60 μm) 3 m from the target area, the coverage of water-sensitive paper with water was 24.9 ± 3.7%, compared to 22.5 ± 1.9% for DRA1, i.e. about 10% lower. For DRA2, the coverage was 21.1 ± 2.6% (about 15% lower compared to water), and for DRA3–19.2 ± 2.0% (about 23% lower).

3. The results have shown that the downwind drift of sprayed liquid depends on the atomization of the droplets in the robot’s jet spraying system, which determines the droplet size (VMD_preset). In the case of a gentle lateral wind (v = 2 m s⁻¹), the atomization of the sprayed droplets in the robotic spraying system was set to VMD_preset=60 μm, and all DRA solutions and water drifted to L = 7.5 m from the spray target. By decreasing the atomization level to VMD_preset=80 μm, it can be observed that only water droplets were detected at L = 7.5 m from the target area, while the sprayed solutions of DRAs only drifted up to 5 m away from the target area. By reducing the level of droplet atomization in the robotic spraying system, i.e. by increasing the VMD_preset, the diameter of the sprayed droplets increased, and it is likely that a higher proportion of larger droplets formed in the spray stream, which influenced the lower drift.

4. To summarize the results of the study, the following management measures can be used to control droplet drift using a robotic jet spraying system, in order of importance: lateral wind velocity, selection of the level of droplet atomization, and the use of chemical solution additives (DRAs). At low lateral winds (2–4 m s⁻¹), all investigated DRA solutions were significantly more effective, however, there were differences between them. Moreover, the difference between the effectiveness of DRA solutions decreases with increasing lateral wind velocity from 2 to 10 m s⁻¹. The effectiveness of DRA also depends on the level of atomization of the droplets: the smaller the droplets are sprayed (VMD_preset=60–80 μm), the lower the effectiveness of DRA (spray drift can be reduced by about 2.5-fold), while spraying larger droplets (VMD_preset=100–120 μm) with DRA reduces drift by about 3.5-fold (v = 4 m s⁻¹). However, droplet atomization does not have such a significant effect on spray drift when droplets are affected by strong lateral winds (v = 10 m s⁻¹), as DRA can reduce the drift by approximately 1.5-fold for both very fine droplets and fine droplets.

5. Future research should build upon findings of this study by evaluating robotic spraying systems in diverse agricultural environments, including various crop types and topographical conditions, to assess how these factors influence spray drift and the effectiveness of management strategies. Further work could also explore the long-term applicability and economic implications of these drift control measures in practical farming scenarios.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by T.J., A.K., D.S., A.A., M.D. and J.B. The first draft of the manuscript was written by T.J., A.K., D.S. and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This study received funding from the Ministry of Education, Science and Sports of the Republic of Lithuania and Research Council of Lithuania (LMTLT) under the Program ‘University Excellence Initiative’ Project ‘Development of the Bioeconomy Research Center of Excellence’ (BioTEC), agreement No. S-A-UEI-23-14.

Data availability

The authors declare that the data supporting the findings of this study are available within the paper and its supplementary material file.

Declarations

Competing interests

The authors declare that they have no conflict of interest.