Institut für Agrartechnik

Permanent URI for this collectionhttps://hohpublica.uni-hohenheim.de/handle/123456789/19

Browse

Now showing 1 - 20 of 184

Mechanisms to alleviate over-generalization in XCS for continuous-valued input spaces
(2022) Wagner, Alexander R. M.; Stein, Anthony
In the field of rule-based approaches to Machine Learning , the XCS classifier system (XCS) is a well-known representative of the learning classifier systems family. By using a genetic algorithm (GA), the XCS aims at forming rules or so-called classifiers which are as general as possible to achieve an optimal performance level. A too high generalization pressure may lead to over-general classifiers degrading the performance of XCS. To date, no method exists for XCS for real-valued input spaces (XCSR) and XCS for function approximation (XCSF) to handle over-general classifiers ensuring an accurate population. The Absumption mechanism and the Specify operator, both developed for XCS with binary inputs, provide a promising basis for over-generality handling in XCSR and XCSF. This paper introduces adapted versions of Absumption and Specify by proposing different identification and specialization strategies for the application in XCSR and XCSF. To determine their potential, the adapted techniques are evaluated in different classification problems, i.e., common benchmarks and real-world data from the agricultural domain, in a multi-step problem as well as different regression tasks. Our experimental results show that the application of these techniques leads to significant improvements of the accuracy of the generated classifier population in the applied benchmarks, data sets, multi-step problems and regression tasks, especially when they tend to form over-general classifiers. Furthermore, considering the working principle of the proposed techniques, the intended decrease in overall classifier generality can be confirmed.
Multi-crop early detection of spider mite damage using hyperspectral data and XGBoost
(2026) Mandrapa, Boris; Spohrer, Klaus; Wuttke, Dominik; Ruttensperger, Ute; Dieckhoff, Christine; Müller, Joachim
The two-spotted spider mite is a globally significant pest affecting over 150 crop species, including cucumbers and strawberries. Its feeding activity leads to chlorophyll degradation and physiological changes in leaf tissue, which alter spectral reflectance properties and enable image-based detection. In this study, hyperspectral imaging (HSI) under controlled conditions was used to classify healthy and spider mite-infested leaves of cucumber and strawberry plants, including asymptomatic infested leaves. Spectral data were analyzed and classified with three supervised machine learning algorithms built on extreme gradient boosting (XGBoost) models. The study had three objectives: (1) to assess the ability of XGBoost to classify multiple infestation states, (2) to evaluate model performance with a reduced set of effective wavelengths, and (3) to determine whether infestation across both crops can be classified using a single, merged model. Using all wavelengths, results showed that classification accuracy was 93 % for cucumber leaves, 84 % for strawberry leaves, and 87 % when combined. With five most effective wavelengths, classification accuracy reached 70 % for cucumber leaves, 65 % for strawberry leaves, and 65 % for cucumber and strawberry leaves combined. The most effective wavelengths were consistently selected from the red-edge and near-infrared (NIR) spectral regions, which highlights their importance for early detection. To the best of our knowledge, this is the first known study to successfully apply a combined machine learning model for early spider mite detection across two different crop species using hyperspectral data under controlled conditions. The results show the potential of machine learning for multi-crop pest detection and could lay the groundwork for practical, sensor-based tools in precision agriculture.
A comparative study on the influence of N‐doped primary and secondary hydrochars on the electrochemical performance of biobased carbon electrodes for energy storage application
(2024) Straten, Jan Willem; Alhnidi, Muhammad‐Jamal; Mustaka, Erlir; Alchoumari, Ghassan; Jung, Dennis; Kruse, Andrea
The hydrothermal carbonization (HTC) technique and subsequent pyrolysis were applied. Herein, two different types of biobased feedstocks, sucrose (Suc) and Miscanthus (Mis), were chosen. Urea served as N precursor for the in situ doping. HTC of Suc and Mis with urea leads to N‐doped hydrochars (N‐HCs). Suc and Mis decompose in different, complex degradation pathways, which leads to the emergence of N‐HCs consisting of distinct ratios of N‐containing primary chars (N‐PCs) and secondary chars (N‐SCs). After pyrolysis, maximum N contents of 5.6 wt % and 4.8 wt % of the N‐doped pyrolyzed hydrochar (N‐PHC) electrodes from Suc and Mis, respectively were reached. The role of PC and SC formation on the impact of N‐doping in the context of physicochemical properties of the N‐PHCs was compared with each other. In this study, they were tested with respect to the influence of N‐PCs and N‐SCs on their electrochemical performance in energy storage application. It turned out that the electrical conductivity (EC) and specific capacitances increased. Highest EC value of 129.1 S ⋅ cm −1 was obtained with N‐PHC based on Suc. Simultaneously, enhanced average specific capacitances of 47.0 F ⋅ g −1 at 5 mV ⋅ s −1 and 3.5 F ⋅ g −1 at 100 mV ⋅ s −1 are ascertained with N‐PHCs from Suc and Mis, respectively.
Multivariate analysis for balancing electrical energy requirement, throughput and oil recovery in mechanical extraction of peanut oil
(2026) Bonzi, Wiomou Joévin; Romuli, Sebastian; Nounagnon, Bignon Stéphanie; Meissner, Klaus; Müller, Joachim
Mechanical pressing stands out as a widely adopted method for plant oil extraction in rural areas due to its simplicity and lower initial investment. This process requires seed conditioning and adjusted pressing parameters to enhance oil recovery. Previous studies have focused primarily on optimising oil recovery. In a sustainable approach, additional parameters such as specific energy and throughput should be considered due to their impact on production costs. This study aims to balance optimal oil recovery, energy efficiency, and economic viability in peanut oil mechanical extraction. Employing a Box Behnken design, the impact of shell ratio (5%, 10%, and 15%), steaming duration (10 min, 20 min, and 30 min), and rotational speed (20 rpm, 45 rpm, and 70 rpm) on specific energy of pressing stage, oil recovery, and throughput were investigated. Results show specific energy ranging from 50.7 Wh kg−1 to 96.3 Wh kg−1 and oil recovery ranging from 83.2% to 91.0%. Both responses were predominantly influenced by rotational speed. Specifically, lower rotational speed led to increased specific energy and higher oil recovery. A desirability function was introduced to perform multivariate optimisation, assigning response importance based on their influence on oil production cost. The results revealed an optimum point at 70 rpm, 5% shell ratio, and 20 min steaming duration, corresponding to a desirability value of 0.61. This optimum corresponds to a minimised specific energy (49.2 Wh kg−1) and maximised throughput (10.9 kg h−1), while providing an acceptable oil recovery (86.2%). This research provides valuable insights into optimising peanut oil mechanical extraction, considering energy efficiency and economic viability.
Biogas potential of agriculture
(2022) Nurgaliev, Timur; Müller, Joachim; Koshelev, Valery
The purpose of this research is to evaluate the biogas potential of agriculture in the typical Russian region. The design of this study was completed using the main kinds of agricultural production of the Tambov region as the feedstocks for biogas production. Average amounts of the feedstocks were calculated on the base of data for the period 2009–2018. The quantities and revenues of electricity, heat, and biofertilizers from biogas produced from various substrates were estimated and mapped for each of the twenty-three municipal districts of the region. Results revealed an average total monetary biogas potential of 88.52 × 10 9 RUB for the Tambov region per year, where 75.43% are provided by electricity and heat energy and the remaining 24.57%—by biofertilizers, therefore, biogas potential of the Tambov region is comparable with biogas potential of a European country. Such feedstocks as sunflower silage, cereal grain, and cereal straw were defined as the most attractive substrates in the region. At the same time, the most of feedstocks being the main farmers’ commodity production are debatable to be used as substrates; as for Russian farmers, biogas production is a new and not well-known technology. Nevertheless, the developed calculation method can now be applied by local authorities of the Tambov region and other regions of the Russian Federation as the base to develop the biogas sector in the most promising areas by supporting farmers and business structures and attracting investments in biogas technology.
Aqueous Kabachnik -Fields synthesis of HMF-functionalized chitosan for thermally-stable LBL coatings on cotton fabrics
(2026) Checa, M.; Arauzo, P.J.; Kruse, A.
This study explores the Kabachnik-Fields (KF) reaction between chitosan, diethyl phosphite (DEP) and biorefinery-derived 5-hydroxymethylfurfural (HMF) to produce chitosan-diethyl phosphite-HMF (CDH) derivatives for thermally-stable coatings. The effect of reaction temperature on the structure, phosphorus incorporation and thermal properties of the CDH-RT (room temperature), CDH-60 (60 °C), CDH-80 (80 °C) and CDH-100 (100 °C) products was investigated using FTIR, elemental analysis, ICPOES and thermogravimetric analysis (TGA) in air. KF modification increased chitosan's initial thermal stability (T₅: 65 to 84-104 °C), promoted early char formation through an intumescent mechanism and enhanced residue stability at 930 °C (max. 32.4 wt %). To assess LBL coating compatibility, CDH products were applied to cotton fabrics (3 bilayers with phytic acid). Mass gain depended on KF temperature, with cotton-CDH-60/PA showing optimal incorporation (34.4 %). Synergy factors calculated from normalized TGA residues confirmed positive char-promoting effects for all CDH formulations (cotton-CDH-60/PA: 226.6 %). These results establish HMF-chitosan aminophosphonates as a promising platform for sustainable, thermally-protective cotton fabric coatings.
B,N‐doped activated carbon‐based electrodes from potato peels for energy storage applications
(2025) Straten, Jan Willem; Alhnidi, Muhammad‐Jamal; Alchoumari, Ghassan; Sangam, Krishna; Kruse, Andrea
Potato peels (PPs) as waste biomass were selected as the biobased carbon source for this study, using urea as N precursor and boron trioxide as B precursor for the “in situ doping” via hydrothermal carbonization (HTC). During HTC, the feedstocks decompose over a wide range of complex chemical degradation mechanisms that finally form single B‐ and N‐ as well as B,N‐co‐doped hydrochars (HCs). Upon chemical ZnCl2 activation, the single B‐doped activated carbon (AC) possessed a maximum B content of 0.2 wt%, whereas co‐doped B,N‐AC had the highest N content of 5.7 wt% with a B content of 0.1 wt%. The influence of single and B,N‐co‐doping on the physical‐chemical material properties of the AC electrodes was analyzed and compared, in combination with its effect on the electrochemical performance for energy storage application. Compared to pristine AC derived from PPs, the B‐doped and B,N‐co‐doped AC depicted increased electrical conductivity (EC) values of 50.3 S ⋅ m−1 and 34.0 S ⋅ m−1, respectively. In addition, the B,N‐co‐doped AC unveiled the highest average specific capacitances of 51.7 F ⋅ g−1 at 100 mV ⋅ s−1 and of 71.9 F ⋅ g−1 at 5 mV ⋅ s−1 outperforming the specific capacitance values of the reference material AC from peat.
Mitigating phytotoxicity of hydrothermal liquefaction hydrochar toward potential agricultural applications
(2026) Batista, Gabriel F.; Kruse, Andrea; Becker, Gero C.
Valorizing hydrothermal liquefaction (HTL) by-products is essential to improve process sustainability and support its industrial-scale implementation. However, the direct agricultural application of HTL-derived hydrochar remains limited due to reported phytotoxic effects. By studying and mitigating phytotoxicity, this work evaluates the potential suitability for agricultural use of hydrochar, the solid by-product from continuous HTL of a 50/50 wt. % cattle manure and wheat straw mixture at 325 °C, separated with an in-line filter. Phytotoxicity was assessed using seed germination assays with Barley (Hordeum vulgare) and Cress (Lepidium sativum) seeds. Two hydrochar post-treatments, washing (hydrochloric acid and water) and pyrolysis (300 °C and 500 °C), were examined to mitigate hydrochar phytotoxicity. Raw HTL-hydrochar significantly hindered plant growth, reducing the root lengths of barley and cress by 37 % and 70 %, respectively, compared to the control. Water-washed post-treatment eliminated hydrochar phytotoxicity and enhanced Barley root growth by 42 % compared to control at a 15 ton ha⁻¹ application rate, indicating a possible growth-stimulating effect. Pyrolysis also mitigated hydrochar phytotoxic effects, with cress root lengths statistically similar to the control. No uptake of heavy metals by the plants were observed in the germination assays. These results suggest that phytotoxicity originates from water-soluble organic compounds, likely phenols, short-chain organic acids and aldehydes, produced during HTL process and adsorbed in the hydrochar surface. The novelty of this work lies in demonstrating the complete removal of phytotoxicity from HTL hydrochar using technologically mature and scalable post-treatments. Therefore, a barrier to hydrochar valorization is removed, enabling further investigations into agronomic applications. This work contributes to a circular biomass valorization strategy.
Predicting herbage biomass on small‐scale farms by combining sward height with different aggregations of weather data
(2024) Scheurer, Luca; Leukel, Joerg; Zimpel, Tobias; Werner, Jessica; Perdana‐Decker, Sari; Dickhoefer, Uta
Accurate predictions of herbage biomass are important for efficient grazing management. Small‐scale farms face challenges using remote sensing technologies due to insufficient resources. This limitation hinders their ability to develop machine learning‐based prediction models. An alternative is to adopt less expensive measurement methods and readily available data such as weather data. This study aimed to examine how different temporal aggregations of weather data combined with compressed sward height (CSH) affect the prediction performance. We considered weather features based on different numbers of weather variables, statistical functions, weather events, and periods. Between 2019 and 2021, data were collected from 11 organic dairy farms in Germany. Herbage biomass exhibited high variability (coefficient of variation [CV] = 0.65). Weather data were obtained from on‐farm and nearby public stations. Prediction models were learned on a training set ( n = 291) and evaluated on a test set ( n = 125). Random forest models performed better than models based on artificial neural networks and support vector regression. Representing weather data by a single feature for leaf wetness reduced the root mean square error (RMSE) by 12.1% (from 536 to 471 kg DM ha −1 , where DM is dry matter) and increased the R 2 by 0.109 (from 0.518 to 0.627). Adding features based on multiple variables, functions, events, and periods resulted in a further reduction in RMSE by 15.9% ( R 2 = 0.737). Overall, different aggregations of weather data enhanced the accuracy of CSH‐based models. These aggregations do not cause additional effort for data collection and, therefore, should be integrated into CSH‐based models for small‐scale farms.
Evaluation of energetic potential of slaughterhouse waste and its press water obtained by pressure-induced separation via anaerobic digestion
(2024) Yankyera Kusi, Joseph; Empl, Florian; Müller, Ralf; Pelz, Stefan; Poetsch, Jens; Sailer, Gregor; Kirchhof, Rainer; Agyemang Derkyi, Nana Sarfo; Attiogbe, Francis; Siabi, Sarah Elikplim; Jeon, Byong-Hun
Anaerobic digestion has the potential to convert organic waste materials into valuable energy. At the same time, using press water from biomass materials for energy generation while taking advantage of the resulting cake for other purposes is an emerging approach. Therefore, this study aimed to investigate the residual potential expected from a typical biogas feedstock after it has been mechanically separated into liquid and solid phases. Hence, in this study, the rumen contents of ruminants (cow, goat, and sheep) and their proportionate ratios were obtained from an abattoir in Ghana. Resource characterization of the waste samples was carried out in the central laboratory of the HFR, Germany. Anaerobic batch tests for biogas (biomethane) yield determination were set up using the Hohenheim Biogas Yield Test (HBT). The inoculum used was obtained from an inoculum production unit at the Hohenheim University biogas laboratory. The trial involved two different forms of the sample: mixture of rumen contents, press water, and inoculum, each in four (4) replicates. The trial was carried out at a mesophilic temperature of 37 °C. Results obtained over a seventy (70) day period were transformed into biogas yields. Overall, the results show that the current contents are suitable for biogas generation as an option as opposed to the current form of disposal at a refuse dump. However, using these mixtures in their original forms is more technically viable than using press water without further treatment.
From coffee waste to wastewater treatment: optimization of hydrothermal carbonization and H₃PO₄ activation for Cr(VI) adsorption
(2026) Piccoli Miranda de Freitas, Caroline; De Freitas Batista, Gabriel; Dalmolin da Silva, Mariele; Checa Gomez, Manuel; Arauzo, Pablo J.; França da Cunha, Fernando; Kruse, Andrea
Spent coffee grounds (SCG) are an abundant agro-industrial waste, and their valorization as activated carbon (AC) offers a sustainable approach for wastewater treatment and heavy-metal remediation. However, the high energy demand of SCG activation limits large-scale application. Hydrothermal carbonization (HTC) reduces energy consumption and enhances material properties. This study evaluated the performance of activated carbon (AC) derived from SCG via HTC, followed by H₃PO₄ activation for Cr(VI) removal, and compared it with non-activated carbon obtained by HTC and pyrolysis. The results highlight the effect of chemical activation on enhancing surface area, porosity, and adsorption efficiency. The predicted optimal IN was 1624.7 mg·g⁻¹, closely matching the experimental value of 1640.1 ± 15.5 mg·g⁻¹, achieved at 426 °C, 92 min, and a hydrochar-to-H₃PO₄ ratio of 1:1.6. The optimized AC exhibited a maximum adsorption capacity (Qₑ) of 33 ± 1.1 mg·g⁻¹ and 99.4 ± 0.1 % Cr(VI) removal under pH 2, 25 mg·L⁻¹ initial concentration, and 2 g·L⁻¹ adsorbent dose. In contrast, the non-activated carbon presented a lower iodine number (1411 ± 70 mg·g⁻¹) and inferior adsorption performance, confirming the key role of H₃PO₄ activation in improving surface reactivity and adsorption sites. Chemical activation proved essential for improving Cr(VI) adsorption, with the H₃PO₄-AC exhibiting the highest capacity. These results demonstrate the potential of SCG-derived AC as a low-cost adsorbent for heavy-metal-rich industrial effluents, supporting circular economy strategies.
How fluid pseudoplasticity and elasticity affect propeller flows in biogas fermenters
(2024) Kolano, Markus; Ohnmacht, Benjamin; Lemmer, Andreas; Kraume, Matthias
Mixing in biogas fermenters is complex due to the non‐Newtonian rheology of biogenic substrates, which exhibit both pseudoplasticity and elasticity. It is yet unclear how these non‐Newtonian properties affect propeller flows and the mixing behavior in fermenters. Therefore, propeller flows in Newtonian as well as shear‐thinning inelastic and elastic fluids are compared numerically and validated against particle image velocity (PIV) data. Elastic normal stresses lead to an increase of pumping rates in the laminar regime and a suppression of the formation of a propeller jet in the transitional regime. Thus, flow rates are severely overestimated by the inelastic, shear‐thinning model in this regime. The results indicate that elasticity is critical for an accurate modeling of flows of biogenic substrates.
Herstellung von HMF aus Kartoffelschalen
(2025) Limbach, Nadine; Konnerth, Philipp; Kruse, Andrea
Die stoffliche Nutzung von Biomasse zur Herstellung von Plattformchemikalien gewinnt zunehmend an Bedeutung für eine nachhaltigere Chemie. Eine wichtige Verbindung in diesem Bereich ist 5-Hydroxymethylfurfural (HMF), das aus einfachen Zuckern gebildet werden kann. Ziel dieser Arbeit war es, HMF aus dem stärkehaltigen Nebenprodukt der Kartoffelschale zu synthetisieren. Dazu wurden die Einflüsse zweier Mineralsäuren – Schwefelsäure und Salpetersäure – in unterschiedlichen Konzentrationen (1 M, 1,5 M und 2 M) untersucht. Die experimentelle Arbeit bestand aus zwei aufeinanderfolgenden Schritten. Zunächst wurde die Stärke der Kartoffelschalen hydrolytisch aufgeschlossen, um eine möglichst hohe Glucoseausbeute zu erzielen. Im anschließenden Versuch wurde diese Glucose über Isomerisierungs- und Dehydratisierungsschritte zu HMF umgesetzt. Hierfür wurden die Reaktionslösungen auf verschiedene pH-Startwerte (pH 2, pH 2,5 und pH 3) eingestellt. Die Ergebnisse zeigen, dass beide Säuren die Stärkehydrolyse in ähnlicher Weise katalysieren und vergleichbare Ausbeuten an Glucose, Fructose und Zucker-Dimeren bei gleicher Verweilzeit liefern. In der nachfolgenden HMF-Synthese traten jedoch deutliche Unterschiede zwischen den Säuren auf: Schwefelsäure führte zu einer schnelleren Zuckerumwandlung und zu höheren HMF-Ausbeuten bei kürzerer Reaktionszeit. Mit sinkendem pH-Wert stiegen die HMF-Ausbeuten bei beiden Säuren an. Neben HMF entstanden weitere Neben- und Abbauprodukte wie Levulinsäure, Ameisensäure und Huminstoffe. Dabei bildete sich bei Verwendung von Schwefelsäure eine höhere Menge an Huminstoffen als bei Salpetersäure. Insgesamt zeigt sich, dass Schwefelsäure die beteiligten Reaktionen bei gleichem pH-Startwert stärker katalysiert.
Effects of pretreatment with a ball mill on methane yield of horse manure
(2023) Heller, René; Roth, Peter; Hülsemann, Benedikt; Böttinger, Stefan; Lemmer, Andreas; Oechsner, Hans
Lignocellulosic biomass is an abundant organic material, which can be utilised in biogas plants for sustainable production of biogas. Since these substrates usually have high lignin contents and consist of rather elongated particles, a special pretreatment is required for an economical and process-stable utilisation in the biogas plant. The mechanical pretreatment of horse manure was carried out with the prototype of a ball mill at different speeds. The aim of ball milling is to comminute the substrate and disintegrate the lignocellulosic bond. Mechanical pretreatment in the ball mill resulted in a significant increase in specific methane yield of more than 37% in anaerobic batch digestion (up to 243 LCH4 kgVS−1) of horse manure. The kinetics of the methane gas formation process was analysed by a modified Gompertz model fitting and showed a higher methane production potential and maximum daily methane production rate as well as a lower duration of the lag phase after pretreatment at 6 rpm. This was further confirmed by sieve analyses, which showed a significant reduction of particle size compared to the untreated variant. Thus, the use of the ball mill increases the specific methane yield and improves the fermentation of lignocellulosic substrates such as horse manure.
Risk analysis of the biogas project
(2023) Nurgaliev, Timur; Koshelev, Valery; Müller, Joachim
The dynamic model of the biogas project was created with changing parameter values over time and compared to the static model of the same project based on constant values of the same parameters. For the dynamic model, the same methods were used to evaluate the biogas project as for the static model to calculate substrate mix volumes, costs, farm production volumes, number of biogas plant equipment, driers, and other numerical characteristics of the farm. Project risks were evaluated by the sensitivity analysis and Monte Carlo simulation. The study was conducted for four scenarios regarding the substrate mix structure and the possibility of selling electricity on the market. In the scenarios, the scale of the project was determined by the size and structure of agricultural and biogas production. The results have shown that when only wastes are used as substrates, net present values (NPVs) of the project are equal to 29.45 and 56.50 M RUB in dependence on the possibility to sell electricity on the market. At the same time, when the substrate mix is diversified, the project NPVs are equal to 89.17 and 186.68 M RUB depending on the ability to sell all the produced electricity to the common power grid. The results of the sensitivity analysis defined that the values of elasticity coefficients are less than 3.14%. Results of the Monte Carlo simulation have shown a probability distribution of positive NPVs for each scenario. This study was conducted to make recommendations for business and municipalities.
Assessing the capability of YOLO- and transformer-based object detectors for real-time weed detection
(2025) Allmendinger, Alicia; Saltık, Ahmet Oğuz; Peteinatos, Gerassimos G.; Stein, Anthony; Gerhards, Roland
Impact of high-pressure processing on the bioactive compounds of milk - a comprehensive review
(2024) Siddiqui, Shahida Anusha; Khan, Sipper; Bahmid, Nur Alim; Nagdalian, Andrey Ashotovich; Jafari, Seid Mahdi; Castro-Muñoz, Roberto
High-pressure processing (HPP) is a promising alternative to thermal pasteurization. Recent studies highlighted the effectivity of HPP (400–600 MPa and exposure times of 1–5 min) in reducing pathogenic microflora for up to 5 logs. Analysis of modern scientific sources has shown that pressure affects the main components of milk including fat globules, lactose, casein micelles. The behavior of whey proteins under HPP is very important for milk and dairy products. HPP can cause significant changes in the quaternary (> 150 MPa) and tertiary (> 200 MPa) protein structures. At pressures > 400 MPa, they dissolve in the following order: αs2-casein, αs1-casein, k-casein, and β-casein. A similar trend is observed in the processing of whey proteins. HPP can affect the rate of milk fat adhering as cream with increased results at 100–250 MPa with time dependency while decreasing up to 70% at 400–600 MPa. Some studies indicated the lactose influencing casein on HP, with 10% lactose addition in case in suspension before exposing it to 400 MPa for 40 min prevents the formation of large casein micelles. Number of researches has shown that moderate pressures (up to 400 MPa) and mild heating can activate or stabilize milk enzymes. Pressures of 350–400 MPa for 100 min can boost the activity of milk enzymes by up to 140%. This comprehensive and critical review will benefit scientific researchers and industrial experts in the field of HPP treatment of milk and its effect on milk components.
AI-assisted tractor control for secondary tillage
(2025) Boysen, Jonas; Bökle, Sebastian; Stein, Anthony
Modern agricultural machinery requires skilled operators to optimally configure their complex machines, while autonomous machines without operators must already optimize their configuration themselves to achieve optimal performance. During secondary tillage multiple performance measures need to be monitored and maximized: Seedbed quality, area output and fuel consumption. The seedbed quality can be measured with the soil surface roughness coefficient which can be computed with 3D-cameras attached to the machine. For our work, such cameras are mounted in the front and back of a Claas Arion 660 tractor with an attached power harrow seeding combination. The soil-machine response model of our prior work is utilized to model the soil-machine interaction for the training of a reinforcement learning agent and the application of a decision-time planning agent to assist in controlling the working speed of the machine. The control agents are tested in real-world field trials and compared to good professional practice. The decision-time planning agent achieves comparable results to a gold-standard while reaching significantly higher performance in terms of area output (29.1%) and more efficient fuel consumption (8.4%) than a baseline while the reinforcement learning agent performed worse during the field trials. The seedbed quality and field emergence are not showing significant differences between the variants. Further analysis shows that model training and selection for the reinforcement agent could have led to performance loss and models that are performing better in simulation have been trained after the field trials. Furthermore, we analyze the models when tested under the field conditions in the field trials (out-of-distribution) that are different from the field conditions during training data collection. The out-of-distribution testing leads to a reduced performance in terms of rRMSE of the decision-time planning agent and to some extend reward of the reinforcement learning agent compared to in-distribution testing.
Assessing the capability of YOLO- and transformer-based object detectors for real-time weed detection
(2025) Allmendinger, Alicia; Saltık, Ahmet Oğuz; Peteinatos, Gerassimos G.; Stein, Anthony; Gerhards, Roland
Spot spraying represents an efficient and sustainable method for reducing herbicide use in agriculture. Reliable differentiation between crops and weeds, including species-level classification, is essential for real-time application. This study compares state-of-the-art object detection models-YOLOv8, YOLOv9, YOLOv10, and RT-DETR-using 5611 images from 16 plant species. Two datasets were created, dataset 1 with training all 16 species individually and dataset 2 with grouping weeds into monocotyledonous weeds, dicotyledonous weeds, and three chosen crops. Results indicate that all models perform similarly, but YOLOv9s and YOLOv9e, exhibit strong recall (66.58 % and 72.36 %) and mAP50 (73.52 % and 79.86 %), and mAP50-95 (43.82 % and 47.00 %) in dataset 2. RT-DETR-l, excels in precision reaching 82.44 % (dataset 1) and 81.46 % (dataset 2) making it ideal for minimizing false positives. In dataset 2, YOLOv9c attains a precision of 84.76% for dicots and 78.22% recall for Zea mays L.. Inference times highlight smaller YOLO models (YOLOv8n, YOLOv9t, and YOLOv10n) as the fastest, reaching 7.64 ms (dataset 1) on an NVIDIA GeForce RTX 4090 GPU, with CPU inference times increasing significantly. These findings emphasize the trade-off between model size, accuracy, and hardware suitability for real-time agricultural applications.
Development and evaluation of a self-adaptable planting unit for an autonomous planting process of field vegetables
(2024) Lüling, Nils; Straub, Jonas; Stana, Alexander; Brodbeck, Matthias; Reiser, David; Berner, Pirmin; Griepentrog, Hans W.
Today, the number of solutions for automated processes in agriculture is growing rapidly. This is primarily driven by the lack of available and affordable labour, pricing pressures, and regulatory requirements. Vegetable production in particular has a lot of potential for automation, as many process steps, such as planting, are performed partly manually. Fully automated systems for the planting process are characterized by their big size, which is only suitable for large farms. At the same time, these planters typically have a low level of intelligence, which is essential for a fully autonomous planting process performed by autonomous vehicles or robots. The following work therefore deals with the development and construction of a prototype for vegetable planting via a robotic platform. This prototype is designed to meet the requirements of a conventional planter and carry out the planting process automatically using a robotic platform. To ensure a robust robotic planting process, an AI-based control system has been integrated that can detect and adjust the planting quality. For this reason, the planting unit was designed to allow dynamic changes in working depth and furrow width. By dynamically controlling these planting parameters, there is potential for a more sustainable planting process with lower energy requirements. A number of evaluations have been carried out to validate the described characteristics of the prototype planting unit.

Browse

Recent Submissions