Path planning for field robots on sloping farmland - Runoff and erosion mitigation by cross-slope cultivation and in-field agri-environmental measures

Path planning for field robots on sloping farmland Runoff and erosion mitigation by cross-slope cultivation and in-field agri-environmental measures M. Melzer, Leibniz Centre for Agricultural Landscape Research (ZALF), Müncheberg; S. A. Pauli, AGCO GmbH, Marktoberdorf; S. Kopfinger, Bavarian State Research Centre for Agriculture (LfL), Ruhstorf a. d. Rott Abstract The field robot Farmdroid FD20 can be used for mechanical weed control in sugar beets. While mechanical weed control reduces herbicide usage, it can increase susceptibility to soil erosion. Soil erosion can be reduced by cross-slope cultivation (CSC) and aligned vegetative barriers. However, CSC using the FD20 is limited to slope gradients of 10 or 5%, depending on whether the FD20 is operating with an active front wheel or not. Beyond these limits, the risk of inaccuracy and crop damage increases. This is not considered by current path planning tools for automated operation of field robots. The suitability of the FD20 for CSC and vegetative barriers on sloping fields is thus unknown. Therefore, sugar beet fields susceptible to erosion are analyzed to identify candidates for CSC with the FD20 considering slope-related limitations of 5 to 15% and potential locations for in-field raised vegetative barriers (beetle banks) by using a novel terrain-based path planning tool. Further, a workflow for implementing beetle banks in the respective fields is investigated. When an active front wheel is used, 81% of 636 analyzed fields are suitable for CSC, compared to only 4% without it. The implementation of beetle banks is practicable but would benefit from additional interfaces for data exchange between the FD20 and external systems. Decision-making processes for both users and developers of field robots can be supported by the analysis. 1. Introduction Two key objectives of the European agricultural policy are to reduce erosion and the use of pesticides [1]. However, these objectives are incompatible in certain circumstances. The field robot Farmdroid FD20 [2], for instance, reduces herbicides usage through automated hoeing in sugar beet production, but might increase the field’s susceptibility to erosion on sloping farmland [3]. To counteract this, erosion can be reduced by additional mitigation strategies such as cross-slope cultivation (CSC) and aligned in-field vegetative barriers [1]. Machinespecific limitations for correct operations on sloping fields (operational limits) pose an obstacle to CSC [4]. According to the manufacturer's specifications, the FD20 can operate without errors across slopes with a gradient of up to 5%, or up to 10% when using an active front wheel [2]. The suitability of the FD20 for implementing CSC on individual fields is unknown because of the described limitations and a lack of practical experience. This information is important, however, for investment decisions and use of the FD20 to achieve greater sustainability in agriculture. A novel path planning tool for tractors has recently been developed that considers the following criteria: a) the number of ‘turning manoeuvres’ required to manage a field along straight waylines, b) the optimal orientation of waylines for CSC (‘wayline gradient’), c) machine-specific limitations for operations across the slope (‘cross-slope’) [1]. The tool also identifies suitable locations for in-field and raised vegetative barriers (beetle banks) that retain water in the field to reduce erosion. Building on this tool, the present analysis aims to i) determine suitable fields for CSC in sugar beets managed with the FD20, ii) identify suitable locations for beetle banks in this set of fields, and iii) demonstrate which procedures, technologies, and data are required for a practical implementation and integration of beetle banks in fields managed by the FD20. 2. Material and Methods A region in southern Germany (central coordinates: 48°45'41.4"N 12°35'43.5"E), comprising 636 fields used for sugar beet cultivation in the years from 2019 to 2022, was selected based on IACS 1 data and a digital elevation model (DEM) indicating heterogeneous terrain susceptible to erosion. The abovementioned path planning tool 2 was used to calculate five wayline sets oriented along the major edges of each field followed by a multiple criteria decision analysis (MCDA) to select the set with the lowest DEM-based contour deviation (wayline gradient) provided that the number of turning manoeuvres and the cross-slope are within reasonable limits to guarantee normal field operations. The field robot’s maximum operability during CSC along the selected waylines was tested against eleven operational limits (5 to 15%), thus including the official limits for an FD20 with or without active front wheel, respectively. To achieve this, the gradients of 2-m-long transects were calculated at 5-m intervals along the waylines, representing the position of the field robot’s axle during CSC (Figure 1). 1 IACS (integrated administration and control system) of the European Union provides field geometries and information on annual crops. 2 Automated tool and user guide available via Zenodo: https://doi.org/10.5281/zenodo.14272617 Figure 1: Zoomed section shows transects and related endpoints colored based on slope gradients to reflect the axel position of the FD20 along waylines of a field. The proportion P (%) of transect gradients above each operational limit and the highest transect gradient value (%) were calculated for each field. Eligible fields for CSC under the eleven operational limits were selected respectively based on the following rules: i) proportion P is less than 1% and ii) the highest transect gradient value is less than 30%. Consequently, about 1% of a field’s area with gradients above the operational limit was considered acceptable and assumed not to impair field management when below the maximum gradient value (considered for safety reasons [4]). This allows for outliers (transects with unexpectedly high gradients), which would lead to the unwanted exclusion of a field. Outliers can arise due to errors in the DEM or at the edges of a field if endpoints of transects are in roads or drainage ditches, for example, that do not belong to the managed area of the field. Potential locations for beetle banks were then calculated along the identified waylines [1]. A workflow was developed to implement a beetle bank within a sugar beet field managed by the FD20, while also considering other hardware and software components of agricultural equipment used. 3. Results The number of suitable fields for CSC using the FD20 increases sharply with operational limits raised from 5 to 9% but flattens between 10 and 15% (Figure 2). Under the operational limit of 5% as specified by the manufacturer for CSC using an FD20 without active front wheel, only 4% of 636 analyzed fields can be managed along the determined waylines. In contrast, an FD20 with active front wheel (operational limit of 10%) would be able to manage 81% of the fields analyzed. Figure 2: Fields (%) suitable for CSC depending on the maximum operable slope gradient during CSC, denoted operational limit (%). FD20-limits of 5 and 10% are boldfaced. In total, 96 potential sites for beetle banks with a minimum length of 50 meters were identified. Figure 3 shows the suitability of fields for CSC based on operational limits of 5, 10 and 15% and locations for beetle banks. Section A of the map shows a beetle bank and underlying waylines almost parallel to contour lines. In section B, however, contour lines deviate from the waylines so that the possibility of CSC is impaired. Figure 3: Exemplary fields suitable for FD20 under three operational limits (5, 10, 15%). Section A: potential beetle bank placed along determined waylines almost parallel to contour lines. Section B: The orientation of waylines differs clearly from contour lines. 0 20 40 60 80 100 5 6 7 8 9 10 11 12 13 14 15 suitable fields for CSC (%) operational limit = maximum operable slope gradient (%) A workflow was identified to integrate a proposed beetle bank into the field management of sugar beets by considering other relevant machinery and processing steps: i) waylines and a proposed location of a beetle bank are transferred to a tractor with GPS-steering system to integrate beetle bank setup in the ploughing and seedbed preparation processes, ii) corner points of field and beetle bank are mapped using the FD20 field setup tool 3 , iii) a path planning tool integrated to the FD20 mobile app provides waylines calculated parallel to a mapped field edge selected by the user, while considering the beetle bank as an obstacle, iv) these waylines are transferred to the FD20 for automated sugar beet sowing and hoeing, v) the initial waylines and position of the beetle bank can be transferred to a GPS-steering system of a beet harvester for precise operation. 4. Discussion Straight waylines deviate from the ideal alignment across the slope when the slope direction varies within a field (Figure 3). The effect of CSC to reduce erosion may thus be impaired compared to contour farming. Field-specific effects on erosion should be quantified by suitable models and provided to farmers for accurate decision making considering additional costs caused by CSC, but also the option of beetle banks in addition to CSC. Path planning requires consideration of machines’ individual working widths while beetle banks are intended to permanently remain on the field [1]. Therefore, the management of subsequent crops and respective equipment needs to be considered as well. An interface or export function would thus be useful to share field boundaries mapped by the FD20 field setup tool with other systems. The calculation of waylines and beetle banks (see first step of workflow) could then be performed independent of IACS data, that might be inaccurate. This would ensure exact matching with the waylines determined by the FD20 path planning tool as described in the third step of the workflow. A direct integration of the novel path planning tool into the FD20 mobile app would further simplify the workflow and its applicability, for instance, to set variables such as headland, working width, and operational limits individually. Farmers’ decision-making for (or against) an FD20 could be supported by showing which fields can realistically be managed with the FD20. The transect-based analysis would allow to identify operational zones of a field based on specified limits (Figure 1). Zones ineligible for the FD20 could instead be used for agri-environmental measures such as grassed buffer strips. This would result in additional erosion control on potential hotspots while the remaining field can be managed by the FD20. 3 Portable RTK GPS antenna that is manually placed on the corner points of fields and permanent obstacles to record and transfer coordinates to the FD20 mobile app. The analysis allows manufacturers to identify eligible regions for their products to optimize marketing. Conversely, technical improvements could be targeted based on region-specific minimum operational requirements. A major improvement for operations with FD20 has already been achieved by providing an active front wheel. However, the operational limits given by the manufacturer refer to optimal conditions on the field [2]. In poor conditions, less fields would be suitable. In field trials, the actual limits could be investigated under real conditions and compared with the results of the presented method for calibration by integrating additional parameters such as soil type and soil moisture to the analysis tool and by adjusting the thresholds (1%, 30%) of applied rules. 5. Conclusion The suitability of the FD20 for CSC is highly dependent on its front wheel configuration. The novel terrain-based path planning tool effectively identifies both CSC-compatible fields and feasible locations for beetle banks. Improved interoperability between the FD20 and external systems would improve applicability to beetle bank implementation. The tool supports not only farmer decision-making and site-specific erosion mitigation but also provides manufacturers with actionable insights for product development and regional marketing. However, the applied threshold values and actual operational limits require validation through planned field trials. Ultimately, combining automated mechanical weeding with strategically planned mitigation measures can contribute to reconciling pesticide reduction and erosion control goals in European agricultural policy. Acknowledgement: This work was part of the Project DAKIS, grant ID: FKZ 031B1524A, financed by the German Federal Ministry of Research, Technology and Space (BMFTR). 7. References [1] Melzer, M.; Spykman, O.; Bellingrath-Kimura, S.D.: Beetle bank-positioning on sloped farmland to promote water retention and biodiversity in farm management information systems for agri-environmental schemes. Biological Conservation. 302 (110999), 2025. [2] Product sheet of Farmdroid FD20 2025 v2.5. https://farmdroid.com/product-sheet/ [3] Kunz, C.; Schröllkamp, C.; Koch, H.-J.; Eßer, C.; Schulze Lammers, P.; Gerhards, R.: Potentials of post-emergent mechanical weed control in sugar beet to reduce herbicide inputs. Agricultural engineering.Eu, 70 (3), 67–81. 2015. [4] Estler, M.; Pfahler, K.: Einfluß der Hangneigung auf den Wert landwirtschaftlicher Grundstücke. Materialien zur Flurbereinigung - Heft 8. 1985.