Introduce and Evaluate an Unmanned Ground Sprayer for Vineyards and Orchards

This often leads to either the overuse of chemicals or inadequate coverage, depending heavily on the operator's expertise. The development of automated technologies in precision agriculture presents the opportunity for unmanned systems to deliver more precise crop protection. In this study, we conducted a series of tests on an unmanned spraying device to assess its effectiveness in controlling the targeted area and achieving optimal spray coverage.

How It Works

The article examines the XAG R150 Unmanned Ground Vehicle Spray Version described on the manufacturer's website. This device includes two main sections: a self-driving unit that uses RTK navigation to autonomously follow a pre-planned path from a smartphone app and a spray unit for precise application. The focus of the article is on the spray unit's performance. It is designed with the ability to adjust its vertical pitch angle to a maximum of 200° and its horizontal heading angle to a maximum of 290° (Figure 1). The unit offers variable flow rates ranging from 2 to 4 L/min and allows for the selection of atomization levels from 1 to 6 to achieve desired fineness outcomes.

Figure 1. The sprayer can tilt up and down as much as 200º and turn side to side up to 290º. Image courtesy of XAG.

Analysis Before Test

The sprayer's effectiveness was tested in both vineyard and apple orchard settings. In the vineyard, the grapevines had been trained to grow in a vertical shoot positioning system, while the apple trees were cultivated in a vertical system. As the vehicle traversed the rows at a steady speed, the sprayer's vertical swing was the primary method for covering the expansive area of the canopy. In these scenarios, swinging the sprayer horizontally was generally not required. It's important for the sprayer to have a precise vertical swing angle to ensure that the spray reaches all areas of the canopy effectively, minimizing any spray that goes off target. To optimize coverage, the range of vertical movement should be set based on the height of the canopy and the distance between the rows.

The heading angle influences how far the droplets must travel to reach the canopy. As they travel over longer distances, the droplets naturally disperse, increasing the coverage area but also losing momentum. Therefore, it's crucial to adjust the heading angle to suit the specific geometrical characteristics of the vineyard or orchard, such as canopy height and row spacing. For the vineyards, the heading angle was adjusted to 45º to accommodate the dimensions of the grapevines (Figure 2); in the apple orchard, the heading angle was set to 0º, meaning the sprayers were aligned perpendicular to the rows of the canopy (Figure 3). We crafted a mathematical model that uses the heading angle, canopy height, and row spacing as inputs to calculate the optimal vertical swing range. This model is designed to ensure that spraying is confined to the canopy area, enhancing precision and minimizing waste.

Figure 2. In the vineyard test, the heading angle was adjusted to 45º to accommodate the dimensions of the grapevines.

Figure 3. In the apple orchard, the heading angle was set to 0º, meaning the sprayers were aligned perpendicular to the rows of the canopy.

Building on the previous analysis and settings, several additional operational parameters were identified that impact spray coverage, namely operation speed and flow rate. These factors were systematically examined in a series of experiments to determine their effects.

Vineyard Test

We tested a total of 9 combinations of operation speeds (0.2, 0.4, and 0.6 m/s) and flow rates (2, 3, and 4 L/min). For each combination, three different grapevines were selected at random to serve as replicates for the treatment. To quantify the spray coverage on each vine, we used water-sensitive paper (WSP) samplers. 9 of these samplers were placed strategically throughout the canopy, dividing it into nine sub-zones. This allowed us to measure not just the total coverage but also to compare the spray distribution across the different areas of the canopy. The atomization level was set as 6 to achieve the best fineness outcomes. The WSPs collected from the experiments were initially digitized using an optical scanner. Subsequently, a specialized software program was employed to calculate the percentage of the area on the sampler that was stained blue/purple (Figure 4), which is indicative of the spray droplet coverage. This metric, referred to as 'coverage,' represents the proportion of the sampler's surface area that received spray droplets.

Figure 4. The area on the sampler that was stained blue/purple is indicative of the spray droplet coverage.

The data indicates that, on average, spray coverage across all tested flow rates tends to decrease as the operational speed increases, as shown in Figure 5(a). A slower operational speed allows the sprayer more time to target different parts of the canopy, resulting in higher coverage. Additionally, increasing the flow rate tends to enhance the average coverage across all operational speeds, as shown in Figure 5(b). The users are able to choose their preferred flow rate setting to match specific spraying needs.

Figure 5. (a) Spray coverage tends to decrease as the operational speed increases; (b) increasing the flow rate tends to enhance the average coverage.

The findings reveal variability in spray coverage across different areas of the grapevine canopy. Figure 6 illustrates this, where 9 WSPs were positioned from the top-left to the bottom-right of the canopy and labeled 1 through 9. The three central zones (WSP 4, 5, and 6) exhibited more coverage compared to the others. This increased coverage in the middle zones is likely due to these areas being more directly in the path of the spray as the sprayer moves vertically. Variations in coverage among zones at the same height are more likely a result of the small size of the WSPs relative to the canopy area. This suggests that the differences in coverage could be attributed to whether the WSPs were directly hit by the stream of water droplets, where randomness occurred.

Figure 6. Variability in spray coverage across different areas of the grapevine canopy was found.

Apple Orchard Test

Drawing from the results of the vineyard trials, the subsequent testing in an apple orchard was streamlined. A consistent flow rate of 3 L/min was employed, and two operational speeds (0.2 and 0.4 m/s) were evaluated. For each operational speed, three trees were randomly chosen to serve as replicates. Given the distinct structure of apple canopies as opposed to grapevines, the placement of the WSPs was adjusted accordingly: 8 WSPs were arranged throughout the canopy, segmenting the apple canopy into eight sub-zones for evaluation (Figure 8).

Figure 7 demonstrates that operation speed has a notable impact on average spray coverage. In Figure 8, analysis of coverage variability within different canopy zones revealed that the middle zones (3, 6, and 7) generally received higher coverage compared to other areas. Zone 2, however, had unexpectedly low coverage, which does not align with our initial hypothesis. This anomaly could potentially be attributed to the misplacement or reorientation of the WSP in that zone due to air movement during spraying.

Figure 7. Operation speed has a notable impact on average spray coverage.

Figure 8. Spray coverage varied across different sections of the canopy.

Summary

Field trials in vineyards and orchards, utilizing varying operational speeds and flow rates assessed through water-sensitive paper samplers, revealed that lower speeds and increased flow rates lead to better spray coverage. The design of the sprayer accommodates user preferences, allowing adjustments in flow rates and swing ranges tailored to the specific dimensions of the plant canopy. Additionally, its autonomous navigation unit can follow a preset path at speeds chosen by the user. Collectively, these features demonstrate a significant advancement in achieving efficient and precise crop protection.

Acknowledgment: This project was supported by a USDA ARS grant under the agreement of #58-5082-1-013.