Fruit pose is an important trait for robotic harvesting that provides not only control information for the robot arm but also indicates whether the fruit can be harvested. A pair of fruit and pedicel has been frequently used as the target object for estimating the fruit pose, and the pose estimation of multiple fruit-pedicel pairs within an image is also being focused on in current research to determine the targets to be harvested efficiently. The multi-fruit pose estimation is performed based on accurate object detection that requires a large amount of training data for feature representation of the object, especially pedicel, in complex background; however, there are spatio-temporal constraints in collecting various fruit-pedicel images, and precise labelling is a laborious task. Easier feature representation for localization can reduce model complexity and data-scale dependency. In this study, multi-fruit pose estimation of oriental melon was proposed, which includes a weakly fruit localization that reduces labelling effort while providing the fruit region containing the essential information for estimation of fruit-pedicel pose. The framework consists of a weak supervision-based localization and pose estimation models connected in sequence. The fruit regions are simultaneously and approximately localized, and each of these regions is then fed sequentially to the pose estimation. The proposed method has the advantage of being able to use a smaller model than existing models, which allows for achieving performance even with a small data scale and also improving real-time performance. The results showed that, compared to YOLO-pose models, our method achieved up to 0.23 higher PDK scores and exhibited an inference speed improvement of up to 8 fps. Our method minimized the effort required for localization and improved speed while demonstrating performance comparable to existing studies.

Fruit pose is an important trait for robotic harvesting that provides not only control information for the robot arm but also indicates whether the fruit can be harvested. A pair of fruit and pedicel has been frequently used as the target object for estimating the fruit pose, and the pose estimation of multiple fruit-pedicel pairs within an image is also being focused on in current research to determine the targets to be harvested efficiently. The multi-fruit pose estimation is performed based on accurate object detection that requires a large amount of training data for feature representation of the object, especially pedicel, in complex background; however, there are spatio-temporal constraints in collecting various fruit-pedicel images, and precise labelling is a laborious task. Easier feature representation for localization can reduce model complexity and data-scale dependency. In this study, multi-fruit pose estimation of oriental melon was proposed, which includes a weakly fruit localization that reduces labelling effort while providing the fruit region containing the essential information for estimation of fruit-pedicel pose. The framework consists of a weak supervision-based localization and pose estimation models connected in sequence. The fruit regions are simultaneously and approximately localized, and each of these regions is then fed sequentially to the pose estimation. The proposed method has the advantage of being able to use a smaller model than existing models, which allows for achieving performance even with a small data scale and also improving real-time performance. The results showed that, compared to YOLO-pose models, our method achieved up to 0.23 higher PDK scores and exhibited an inference speed improvement of up to 8 fps. Our method minimized the effort required for localization and improved speed while demonstrating performance comparable to existing studies.

In this paper, the noise and vibration characteristics of axial flux permanent magnet synchronous motors were compared and analyzed according to different pole-slot combinations using an electromagnetic-structural coupled analysis. Under load conditions, finite element analysis was performed to evaluate torque ripple and the axial electromagnetic force imbalance acting on the stator tooth surface for each configuration. Modal analysis was then conducted to identify the natural frequencies of the stator. The electromagnetic forces were applied to a coupled electromagnetic-structural simulation to analyze their effects on vibration and noise responses in the frequency domain. Finally, the NVH characteristics of each pole-slot combination were compared using noise levels and waterfall diagrams at various operating speeds.

This study investigated the energy characteristics of a 100 hp-class agricultural tractor under representative agricultural operations using field-measured workload data and a scenario-based energy analysis framework. Six representative operations, including moldboard plowing, subsoiling, rotary tillage, baler operation, transport operation, and loader operation, were analyzed to evaluate operation-level energy characteristics under realistic agricultural conditions. The results showed that tractor energy demand varied substantially depending on the workload characteristics of each task. Traction-intensive operations exhibited sustained high-power demand, whereas rotary tillage showed the highest specific energy consumption (SEC) because of continuous power take-off (PTO)-driven operation. Utility-oriented operations generally exhibited lower SEC values than traction- and PTO-intensive operations. Scenario-based analysis further demonstrated that workload composition strongly influenced annual energy use (AEU). The traction-intensive scenario exhibited the highest AEU, whereas the utility-oriented scenario showed the lowest cumulative annual energy demand. Sensitivity analysis revealed that annual operating hours had an approximately proportional influence on AEU across all scenarios. The proposed framework provides a practical basis for evaluating workload-dependent tractor energy characteristics and offers useful insights for future agricultural tractor electrification strategies.

Understanding soil-metal adhesion under varying moisture and texture conditions is essential for predicting soil-tool interactions in agricultural machinery. However, despite its importance, research that directly quantifies soil adhesion and provides physically based parameters for modeling remains limited, particularly across the diverse conditions encountered in field operations. In this study, we experimentally measured soil-metal adhesion across multiple penetration speeds and integrated the results with discrete element method (DEM) simulations based on the Edinburgh Elasto-Plastic Adhesion (EEPA) model. Adhesion tests were conducted for three soil textures (sandy loam, sandy clay loam, and loam) and four water contents (10-25%) at penetration speeds of 50 and 500 mm·min⁻¹. Adhesive force increased with water content and peaked near the liquid limit (20-25%), while remaining nearly independent of penetration rate (< 5% variation). DEM calibration showed that surface energy (Δγ) is the dominant parameter governing compressive behavior, whereas the constant pull-off force (f₀) primarily controls adhesive strength. Two-way ANOVA confirmed that these mechanisms operate independently (p > 0.05). The calibrated model achieved R² ≥ 0.93 and RMSE ≤ 0.1 N across all textures, and validation at an intermediate speed of 250 mm·min⁻¹ demonstrated stable predictive performance (R² ≥ 0.93 for compression; R² ≥ 0.98 for adhesion). By linking multi-speed adhesion measurements with a physically based DEM contact model, this work establishes a robust and transferable Δγ-f₀ calibration framework for modeling soil adhesion in cohesive soils.