Conference Agenda

Retrieving crop nitrogen (N) status is crucial for sustainable N management in agriculture. Nowadays, remote sensing and machine learning (ML) have made it feasible to non-destructively estimate crop N status. However, most ML models heavily rely on labeled training samples, potentially compromising their robustness under changing conditions. To address these issues, this study proposed a physics-guided ML approach that integrates physics-based radiative transfer models and empirical ML models. The proposed method was validated using hyperspectral data collected at both leaf and canopy levels. At the leaf level, we employed an ASD LabSpec spectrometer to acquire leaf VIS-NIR-SWIR spectral reflectance, followed by manual measurements of leaf chlorophyll a+b concentration. At the canopy level, VIS-NIR hyperspectral images were captured using a hyperspectral camera mounted on an Unmanned Aerial Vehicle at various study sites. Concurrently, crop N content was destructively sampled. Compared to the standard ML method, the proposed method demonstrated superior performance, particularly when only a small fraction (less than 5%) of the total training samples were available at the leaf level. At the canopy level, the proposed method achieved on average higher accuracy than ML and vegetation indices for leaf N content (RRMSE ranging from 10.08% to 10.84%) and canopy N content (RRMSE ranging from 13.89% to 25.21%) estimation. These findings underscore the benefits of integrating physics-based models and empirical ML models in the proposed approach. This approach offers a promising alternative for accurate estimation of crop N status, with potential applicability to other crops and typical crop traits.

Nitrogen has a major physiological impact on grapevines. Therefore, it is crucial to accurately analyze the nitrogen content of leaves since it directly affects both the productivity of the vineyard and the quality of wine produced.

For this research hyperspectral images of grapevine leaves were taken during two vital growth phases—bloom and veraison—from a ground-based hyperspectral sensor. There were two cultivars Chardonnay and Pinot Noir in two different experiment sites in Oregon state. The study employs a comprehensive approach combining hyperspectral imaging data (spectral features) and chemical analysis (measured nitrogen content) of each grapevine leaf to assess nitrogen content in grapevine leaves. The data is processed using machine learning techniques, specifically ensemble feature selection and partial least square regression feature selection method. Both feature selection methods select a set of features (Wavelength bands) in visible and near infra-red range to assess nitrogen in grapevine leaves.

The results of Chardonnay showed a good predictive performance with a coefficient of determination (R²) value of 0.68. The Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) are 0.28% and 0.22%, respectively. The results of Pinot Noir showed predictive performance with an R² value of 0.54. The RMSE and MAE are 0.24% and 0.19%, respectively. The nitrogen estimation results showed a reliable degree of accuracy that outperformed conventional chemical analysis methods in terms of time, workforce, and non-destructiveness.

The results highlight that hyperspectral imaging can be used to assess the nitrogen content in grapevine leaves and can improve nitrogen control in agriculture.

Central to success in precision agriculture (PA) is the accurate characterization of the spatiotemporal variability of soil information, including physical and hydrological properties and state variables. Researchers have studied soil information using near-surface geophysical methods such as Ground-penetrating Radar (GPR) and Electromagnetic Induction (EMI). Recently, the integrated GPR-EMI approach has gained attention for providing comprehensive insights into soil investigation while overcoming their intrinsic limitations. This review explores the advancements, challenges, and potential applications of the integrated GPR-EMI technique. GPR offers non-invasive, high-resolution imaging of subsurface and soil water content (SWC), facilitating precise delineation of soil stratigraphy and hydrological features. EMI exploits subsurface soil electrical conductivity variations to infer properties and state variables like soil texture, SWC, and salinity. Integrating these two leads to interactive benefits, enabling a comprehensive understanding of the variability and dynamics inherent in soil. Over the past several years, field experiments were conducted with an integrated GPR-EMI technique for predicting different soil information. Findings reveal that GPR has the potential to calibrate EMI for estimating SWC covering similar sampling volumes. Furthermore, the integrated technique improves the prediction accuracy of soil bulk density non-destructively in the studied site. Integrated GPR-EMI opens challenges like requirements for high-quality data, detailed data processing, and site-specific calibration requirements. Future research should focus on refining integration techniques by incorporating geophysics with petrophysical models, standardizing protocols, advancing data analysis using artificial intelligence and validating findings across diverse agricultural settings to realize the full potential of this integration to support precision agriculture.

This research explores the application of Detectron2 for detecting Colorado potato beetles in high-resolution images from a moving vehicle in agricultural fields. The study addresses challenges like the small size of beetles and distortion from resizing images, which can make them nearly invisible, compounded by the variable backgrounds and lighting of outdoor settings. To mitigate these issues, we implemented an image preprocessing strategy to boost detection performance which depends on cropping 1920x1080 to 640x640 to maintain beetle representation accuracy. Our images were taken using an imaging system mounted on a sprayer boom at 500 mm above the canopy - resulting in a dataset of 331 training and 41 testing images, all while exposing the system to diverse field conditions. We trained two models: one on original resolution images and another on cropped images. Testing showed the cropped image model improved detection rates from 48% to 70%, and F1-Score from 62% to 82%. To further enhance detection accuracy, a transfer learning approach was adopted to overcome the small size of the original dataset. Rather than utilizing a pre-trained model with the open-source images, a model was trained on a dataset of static, 4k images captured through a handheld device, cropped to 640x640. This new model was then fine-tuned with the original, moving vehicle dataset (640x640), and detection accuracy was increased to 85%, with an F1-Score of 92%. Our results offer a promising path forward for spot-spraying technologies, minimizing agrochemical usage while managing pests, in line with the principles of precision agriculture.

Application of manure on agricultural lands may release considerable amounts of air pollutants, including gases, particulate matter, and odors, potentially impacting workers, animals, and community health. The objectives of this research are to quantify and compare fugitive emissions (gases, dust, and odors) from five manure types (swine, beef and dairy cattle, and poultry with and without litter) when applied on the field using six spreading methods. Three application techniques were used for solid manure: two conventional spreaders (horizontal and vertical beaters), and a novel prototype for direct incorporation of manure. The three other options were used for liquid manure: splash plate, dribble bar, and dribble bar with immediate incorporation. Each manure and spreader combination was performed in 3 repetitions, from June to November 2023. Preliminary results of field spreading experiments for swine manure showed that incorporation reduced odors and tended to decrease the concentration of ammonia (NH3). Carbon dioxide (CO2) and methane (CH4) concentrations increased slightly when spreading with a splash plate. Odor intensity was substantially greater with the splash plate compared to the dribble bar with incorporation (approximately 2–6 times). Moreover, the splash plate dispersed more particles during application than other methods. Additional findings on other manure types are currently being analyzed and will be included in the final paper.

In Wisconsin, America’s Dairyland, optimizing liquid manure application is crucial for sustainable and profitable farming. Traditional application methods pose environmental challenges, including nutrient runoff and odour emission. This study presents an innovative liquid manure applicator design, underpinned by extensive use of numerical simulations, specifically Finite Element Analysis (FEA), Discrete Element Method (DEM), and Computational Fluid Dynamics (CFD). These simulations played a pivotal role in every phase of the design process – from initial concept to final validation. They allowed for precise modelling of the structural integrity of applicator and soil and fluid dynamics, facilitating the development of three distinct and effective designs: a sweep injector, a disk injector, and a vertical tillage incorporation toolbar. Numerical simulations were integral in predicting the performance under various design alternations and operational conditions, ensuring resilience against mechanical stresses, and optimizing the environmental footprint. This comprehensive simulation-led approach was instrumental in developing the applicator that stands at the intersection of agricultural efficiency and environmental sustainability. By bridging advanced engineering techniques with practical agricultural needs, the project underscores the transformative potential of numerical simulations in modern agricultural/biosystems engineering.

Slurry and manure from animals are valuable sources of plant nutrients, including nitrogen, phosphorus, and potassium. However, over-fertilization poses a significant environmental challenge, leading to the volatilization of excess nitrogen and contributing to various issues such as air pollution, eutrophication, acidification, and greenhouse gas release. Effectively addressing ammonia emissions during manure spreading is crucial for mitigating these environmental problems. The mitigation of ammonia emissions not only brings environmental benefits but also holds social advantages by reducing the adverse effects of ammonia emissions on human health. Recognizing the growing interest in low-cost measurement methods due to their accessibility, affordability, portability, and user-friendly nature, this research aims to bridge a critical knowledge gap regarding the accuracy of these methods, particularly at low concentrations. To address this gap, the study compares ammonia emissions during manure spreading under controlled conditions, employing two low-cost methods: 1) a novel prototype of a passive flux sampler and 2) a dynamic chamber coupled with an acid trap. The experiment, conducted in four controlled rooms, involved the application of pig slurry to a layer of loamy clay soil (120 x 180 x 12 cm), with and without incorporation. The results not only validate the performance of the new low-cost ammonia sampler under controlled conditions but also enable a comprehensive comparison of the accuracy of each method. This research contributes valuable insights towards developing effective and affordable strategies for managing ammonia emissions during agricultural practices.

Understanding corn stalk cutting behaviors under the impact of disc is critical for designing disc for conservation tillage. In this study, three discs (notched, rippled, and plain disc) were examined on the effects of travel speed on corn stalk cutting behaviors using a stalk-disc-soil interaction model developed by the discrete element method. Soil was modelled using spherical particles. Corn stalk was modelled by bonded spherical particles, forming a solid and breakable model of corn stalk. During the simulation, model corn stalks were placed on soil surface and cut as the disc advanced. Dynamic attributes, including stalk cutting forces, stalk displacements, and soil sinkage under the pressure of stalks were monitored. Comparing the simulation values of stalking cutting force with experiment results in literature, the calibrated model parameters of corn stalk were 2 × 10^9 N m^-1 for bond stiffness, 8 × 10^6 Pa for bond strength, and 0.5 for particle friction. The anticipated results are that the model has a low relative error as comparing with measured in a soil bin test. Among three discs, the rippled disc has the least stalk cutting force, following by the notched disc and plain disc. Increasing the travel speed of discs increases stalk displacements, while reduces stalk cutting forces and soil sinkage. The results demonstrated that the stalk-disc-soil is feasible to obtain dynamic attributes of the interaction between soil, disc, and corn stalks during the process of tillage.

Air jet pumps allow the transportation of agricultural materials with no moving parts in the air stream. They function by using a jet of air to entrain and draw air through the duct, which draws in product and conveys it to its intended destination. This research examined the influence of duct shape on air entrainment in air jet pumps used for agricultural material conveyance. By employing computational fluid dynamics (CFD) modelling alongside experimental validation, various duct shapes were analyzed to determine their impact on air entrainment efficiency. The air jet pump was optimized by maximizing air entrainment for a given mass flow.

Common seed manifold designs involve complex interactions between seeds, air, and machine boundaries that are better understood through computer simulations. One-way coupling of Computational Fluid Dynamics (CFD) and Discrete Element Method (DEM) was applied to model seed dynamic attributes (trajectory, velocity, and force) in an air seeder manifold. Simulated field peas (Pisum sativum) were pneumatically conveyed at a rate of 0.07 kg/s, three air velocities (20, 25, and 30 m/s) and three manifold inclination angles (θ = 0°, 11°, and 22°). Model validation was conducted through experiments with a test bench replicating the simulated conditions. Simulated seed trajectories were evenly distributed within the vertical tube and were not significantly affected by manifold inclination angle or inlet air velocity. Median seed contact force ranged between 0.44 N and 3.52 N and was affected by location within the simulated manifold areas (elbow, vertical tube, and manifold head). Contact force data showed that inlet air velocity was a significant factor in simulated seed contact force, while manifold inclination angle did not have an effect. Model validation results revealed that one-way CFD-DEM coupling can replicate experimental observations related to overall seed distribution patterns (R2=0.90). Increasing inlet air velocity promoted more uniform seed distributions, while increasing the manifold inclination angle had the opposite effect. The proposed method lacked accuracy in determining the actual number of seeds per outlet (RMSE = 120 seeds). However, it is complementary to experimental data. This method could be appropriate in benchmarking multiple seed manifold arrangements relative to a baseline design.

Traditional farming methods for fruits and vegetables often involve labor-intensive tasks such as planting, harvesting, and sorting, which can be time-consuming and expensive. While automation technologies have made significant advancements in other sectors of agriculture, adapting these technologies to the unique requirements of fruits and vegetable production poses challenges. Recent advancements in robotics, computer vision, and machine learning are helping to overcome these challenges. However, there remains a notable gap in the adoption of automation technologies, specifically lack of technical skills among the agricultural workforces. Our research focused on the adaptive modular automation for adaptive transformation. Modular components would easily mount on a self-propelled unit (tractor, ATV, or others) as an auxiliary component or attach to fully autonomous equipment (UGV) to perform the intended operation. In this study, will present the performance intelligent cutting mechanism for autonomous lettuce harvester as a modular automation system. It is essential to maintain the consistent cutting height for supermarket quality products. An electrohydraulic control has been designed for autonomous height adjustment for consistent cut. Performance will be evaluated for control system accuracy and lettuce harvesting accuracy at the lab setting. Lettuce cutting will be evaluated for three different blade types, two different blade speed and two different forward speed. Results will be presented during the conference.

This study focuses on the development of an ISOBUS (i.e. ISO 11783) compatible Electronic Control Unit (ECU) for seamless integration between a boom sprayer and machine vision to control individual nozzles for targeted pest spraying. The ECU, powered by a Raspberry Pi 4B and a CAN transceiver, utilizes the AgIsoStack++ (formerly known as ISOBUS++), an open-source library for ISOBUS communication. Through this library, we register our ECU to the ISOBUS network with a manufacturer ID (0x1C) and all parameters of ISOBUS NAME from a registered and inactive manufacturer under the Agricultural Industry Electronics Foundation (AEF). The developed algorithm is specifically ISOBUS Section Control Function (SC) which manages the individual sections of implements, such as nozzle sections on a sprayer controlled by individual cameras for precise spraying operations. The interactive Virtual Terminal (VT) screen design utilizes another freely available software, ISO-Designer. After successful handshaking messages holding 0x1C as the source or destination ID, the designed VT screen becomes visible. The ECU can adjust and control parameters such as camera and nozzle ratios through the ISOBUS compatible VT. Deployment across three different ISOBUS systems resulted in zero error messages, ensuring seamless network registration. The ECU evaluation reveals a 7.1% busload increase on the ISOBUS. The maximum observed busload over those three systems reached 21.3%, remaining within the acceptable limit of 25%. Despite the increase in busload, the ECU increases robustness by integrating the machine vision to VT and simplifies operation by reducing the need for additional displays and interfaces, thereby enhancing farm efficiency.

Wild blueberries (Vaccinium angustifolium Ait.) are a perennial crop in northeastern North America. Conventional management relies on costly broadcast applications of herbicides to manage weeds. There is an opportunity to reduce agrochemical usage by implementing targeted spray applications with the help of deep learning. A machine vision system for targeting weeds in wild blueberry fields was developed as a retrofit package for commercial boom sprayers. The YOLOv5n convolutional neural network (CNN) was trained to detect hair fescue (Festuca filiformis Pourr.) in images of wild blueberry fields. Cameras with an integrated deep learning processor were used to capture images and detect target weeds in real-time. An ISOBUS individual nozzle control system consisting of addressable nozzle bodies and a rate controller was connected to the cameras using a local area network and a custom-built machine vision node. This machine vision spraying system was installed on a 12-nozzle prototype research sprayer built on a John Deere Gator XUV 825i. Testing occurred in Nova Scotia wild blueberry fields in November and December 2023, then April and May 2024. Future work will involve testing the machine vision system on a 27-nozzle commercial wild blueberry sprayer and testing CNNs for targeting different species of weeds in wild blueberry fields. The machine system is adaptable to other crops by retraining the CNN with new images of the desired target weeds. A machine vision smart sprayer will allow growers to achieve substantial cost savings by selectively applying herbicides based on real-time visual data.

Since 1972, wild blueberry harvesters have relied on 0.86 m wide picking heads to optimize field efficiency while ensuring maximum yields. Recently, Doug Bragg Enterprises introduced a 1.47 m picking head, and the purpose of this study was to compare its field efficiency to the conventional 0.86 m header and hand raking. The experiment used a randomized complete block design with four blocks in each field. Within each block, a 75 m strip was harvested with the 0.86 m and 1.47 m heads and the total harvested berry weights were compared to the harvested area. For hand raking, 10, 1 m by 1 m, randomly positioned quadrats were harvested per 75 m strip. In terms of harvested yield by area, there were no significant differences between the 0.86 m head, the 1.47 m head, or hand raking (p = 0.676). This result is encouraging, considering the 1.47 m picking head can harvest 0.166 ha h-1 while the 0.86 m head can only harvest 0.097 ha h-1 at a typical harvesting speed of 0.31 m s-1. Finally, the tested fields had maximum slopes of 9.4 degrees, making them reasonably flat among Nova Scotian wild blueberry fields, which often range up to 30 degrees of slope. Future research should consider header performance in fields with more uneven topography.

Remote supervision of autonomous agricultural machines involves the interaction of the human supervisor and these autonomous machines through an interface. These interfaces need to provide humans with accessible and effective information to perform their supervisory roles, which include setting tasks, allocating resources, monitoring the execution of tasks, and intervening in cases of emergencies. Various warning methods, specifically visual, auditory and tactile cues have been used to inform remote supervisors about any abnormalities in the autonomous machines. Failure to adequately understand the warning information may lead to system breakdown or crop damage. The present study assessed which warning method (visual, auditory or vibrotactile) is most effective in conveying comprehension of machine malfunctions to remote supervisors of autonomous agricultural machines. Twenty-five (25) participants were recruited and asked to interact with a simulation of an autonomous sprayer. Three (3) unimodal warning methods (visual, auditory and tactile sensory cues) were evaluated. The effectiveness of the warning methods was assessed based on the modality’s ability to convey comprehension (which is a measure of level two situation awareness) as a function of accuracy of the participants’ responses. At this time data analysis is being conducted; we plan to share preliminary results at the conference. This study will inform designs that can enhance performance and minimize hazards experienced by remote supervisors during field operation with the autonomous agricultural machines.

With the rise of semi-autonomous tractors in agriculture, there might be a chance to rethink traditional tractor cab designs. This research involves transforming these cabs into ergonomic office spaces, acknowledging that the operator's focus has shifted away from driving. In the initial phase of the project, the emphasis was on conducting an ergonomic assessment of the existing tractor cab design, specifically concentrating on the factors contributing to discomfort. This examination involved analyzing the duration spent on rearward observation, the sequence and frequency of control usage, and ergonomic triggers leading to discomfort. The investigation into tractor ergonomics encompassed interviews with farmers and design engineers, as well as practical experiences of operating tractors through ride-along. Subsequent research delved into an anthropometric analysis, examining the proportions of the human body, and identifying areas prone to discomfort, with a particular emphasis on hand positioning. The results of the ergonomic analysis have pinpointed several issues, including neck and back pain resulting from frequent backward glances, prolonged exposure to vibrations, restricted legroom, and discomfort arising from complex button layouts. Simultaneously, the findings of the anthropometric analysis are being consolidated to generate recommendations tailored for engineers. These recommendations offer insights into key considerations for enhancing human comfort during tractor operation. The ultimate objective of the project's concluding phase is to integrate the insights gained from both ergonomic evaluations and anthropometric analysis into a conceptual design of a comfortable environment for operators of semi-autonomous tractors, effectively transforming the tractor cab into a workspace resembling of an office setting.