Addressing the issues of missed sowing in pneumatic cotton dibblers, as well as the incompatibility and time lag of current reseeding devices, a self-selective seeding method based on a seed tray with inner and outer dual-ring suction holes was developed. A concave arc-shaped interdigital capacitive sensor is used to collect real-time information on the seeds carried by the outer-ring suction holes. Based on this information, a selective seeding execution mechanism controls the target seed object to achieve selective seeding. A mathematical model of the capacitance output of the concave arc-shaped interdigital capacitive sensor was established, elucidating its mechanism for detecting the seed-carrying status. Using Fluent-Rocky DEM coupled simulations and bench tests, the effects of seed tray rotational speed, chamber negative pressure, and suction hole position distribution parameters on the system's seed-filling, seed-cleaning, and seeding performance were investigated, and the optimal operating parameter combination and suction hole position distribution parameters were determined. Through timing analysis of the selective seeding process, a mathematical model between the servo delay drive time and the angular velocity of the seed tray was established, and a corresponding control strategy was formulated. Bench tests were conducted by artificially blocking suction holes to simulate intermittent missed filling and continuous missed filling. The tests show that the system achieves continuous and synchronous precise reseeding for missed sowing. The average missed sowing detection accuracy and the qualified rate of selective seeding are both above 95.07%. This study provides a reference for research on seeding and reseeding methods for pneumatic dibblers.

In variable-rate maize seeding, optimal seeding rate (OSR) decisions are jointly shaped by multiple interacting field factors, including climate, management, and soil conditions. However, the prevailing two-stage "yield prediction-optimal search" framework struggles to effectively capture the synergistic effects among heterogeneous features and is prone to error propagation and accumulation, thereby compromising decision stability. Meanwhile, the long duration of field trials and strong environmental disturbances lead to limited and unevenly distributed OSR ground-truth samples, further constraining predictive accuracy and generalization. Therefore, it is imperative to develop a multi-source, synergy-driven OSR decision-making approach tailored to small-sample scenarios. To this end, climatic variables, field management data, and soil organic matter (SOM) information covering interannual environmental variability were collected over four consecutive years under real-world maize production conditions. Based on these data, a Dual-dimension Adaptive Gating OSR Decision Network (DA-AGNet) was proposed. DA-AGNet incorporates multi-head self-attention, a dual-dimension weighting module, and a gated weighted aggregation mechanism to enable adaptive weighted fusion of multi-source inputs, thereby mitigating the adverse effects of noise in small-sample settings and enabling direct end-to-end prediction of OSR. To improve training stability under small-sample conditions, a coarse-to-fine yield-guided ladder training architecture (CF-YLTA) was constructed. CF-YLTA leverages a yield prediction model to generate pseudo-OSR data for coarse-stage pretraining of DA-AGNet and to initialize transferable prior weights, and then conducts fine-stage end-to-end training by combining these priors with data-augmented OSR ground-truth samples, thus improving prediction accuracy and generalizability. For yield predictor selection, multiple machine learning algorithms were benchmarked, and gradient boosting decision trees (GBDT) achieved the best performance, with the root mean square error (RMSE) of 360.71 kg/ha, a mean absolute error (MAE) of 248.70 kg/ha, and R2 of 0.88. DA-AGNet trained using CF-YLTA delivered the best OSR prediction on the test set (RMSE = 1646.64 seeds/ ha, MAE = 1247.96 seeds/ha, R2 = 0.74), outperforming conventional approaches overall. Qualitative analysis of the predicted OSR indicated an S-shaped increasing trend as SOM increased, which can be partitioned into a low-OSR plateau (SOM 16-25 g/kg), a rapid-increase region (SOM 25-37.5 g/kg), and a high-OSR plateau (SOM 37.5-44 g/kg), consistent with established agronomic understanding. To evaluate model effectiveness, the variable-rate seeding prescriptions derived from DA-AGNet were compared with constant low-and high-density seeding schemes, and the variable-rate strategy showed a yield advantage. Overall, the proposed OSR decision model provides reliable intelligent decision support for determining maize OSR in variable-rate seeding operations under varying crop growth environments.

To investigate the causes of grain breakage during the corn harvesting, this study designed an electronic corn ear based on internet of things (IoT) technology. The electronic corn ear integrates an ESP32 module, an ultra-wideband (UWB) module, an inertial measurement unit (IMU) module, and a flexible film pressure sensor. By utilizing IoT technology, it can detect various kinematic and dynamic parameters of the corn ears during harvesting. To improve detection accuracy, the working parameters of the electronic corn ear were calibrated one by one. The results show that the average detection error of the static compressive force is ≤ 0.93 N, whit the maximum detection error ≤ 6.49 N, and the average relative error ≤ 3.90%. The average detection error of the dynamic impact force is ≤ 0.72 N, with the maximum detection error ≤ 1.92 N, and the average relative error ≤ 2.80%. The spatial positioning error is < 10 cm, and the average positioning errors along the X-axis, Y-axis, and Z-axis are 4.12 cm, 8.44 cm, and 8.18 cm respectively. Using the calibrated electronic corn ear, a corn threshing experiment was conducted. The kinematic and dynamic parameters such as the motion trajectory, the average dynamic impact force, the average impact frequency, the average static compressive force, and the average threshing time of the electronic corn ear during threshing process were detected. The force of the corn ears at different positions in the threshing device and the variation patterns of the grain breakage rate were also identified. This study provides an application scenario of IoT technology in the field of corn harvest damage detection. It can provide data support for elucidating the mechanisms of corn threshing damage and offer technical support for the detection of damage sources during the harvesting process.

Sowing depth consistency is an important factor affecting maize emergence uniformity and subsequent yield formation, while row-unit downforce directly influences opener penetration stability and the ground load state of the gauge wheels. Existing planters are mostly regulated using mechanical springs or fixed-target downforce control methods, making it difficult to determine an appropriate downforce in real time according to changes in soil conditions and target sowing depth. This can lead to insufficient sowing depth stability under complex field conditions. To address this problem, an intelligent decision-making method and system for row-unit downforce during real-time maize seeding were proposed. Sowing depth, soil moisture content, soil firmness, and target downforce were used as influencing factors, while the qualified sowing depth rate and the coefficient of variation of sowing depth were used as evaluation indicators. Response surface analysis was used to analyse the effects of each factor and their interactions on seeding quality and to screen the key input variables for target downforce decision-making. On this basis, multiple linear regression, random forest, and support vector machine algorithms were used to construct sowing quality prediction models, and model performance was compared using the correlation coefficient, standard deviation, and root mean square error. The results showed that, within the experimental range, sowing depth and soil firmness were the key variables affecting target downforce decision-making, and the multiple linear regression model showed better prediction stability and accuracy. With the objective of improving the qualified sowing depth rate while also reducing the coefficient of variation of sowing depth, a model-recommended target downforce table under different sowing depths and soil firmness conditions was established. An intelligent downforce decision-making system was then developed by integrating the decision-making model with an electro-hydraulic regulation mechanism. Field validation results showed that, compared with the traditional spring-based planter row unit, the intelligent downforce decision-making system increased the qualified sowing depth rate by 3.73%–36.74%, reduced the coefficient of variation of sowing depth by 2.40%–8.55%, and reduced the coefficient of variation of emergence time by 3.18%–11.72% under different treatments. The results indicate that the system can recommend an appropriate target downforce according to target sowing depth and soil firmness, improve maize sowing depth consistency and emergence uniformity, and provide a methodological basis and system support for intelligent row-unit downforce regulation.

High-speed precision sowing improves maize production efficiency and contributes to stable yield formation. However, in centrifugal high-speed precision seed metering devices, seeds frequently contact metering components during high-speed filling and clearing, which may cause seed damage. Existing studies on centrifugal seed metering devices have mainly focused on metering performance, whereas the mechanisms of seed damage and the effects of key operating parameters on seed damage remain poorly understood. To address the above research gap, this study established a mechanism-oriented research framework to systematically characterize maize seed damage during centrifugal high-speed precision seed metering. A calibrated Bonding V2-based maize seed fracture model was incorporated into a discrete element method (DEM) model of the seed metering process, and theoretical analysis, DEM simulations, bench tests, and germination tests were jointly conducted. The results showed that seed damage was mainly caused by shear, compression, and impact. Shear damage was concentrated in the mechanical interference zone between the shaped hole ears and the seed-clearing outlet; compression damage mainly occurs in the constricted channel and narrow-gap regions; and impact damage is associated with high-frequency collisions between seeds and rotating components or the housing. Seed plate rotation speed (SPRS) mainly affected seed kinetic energy and impact intensity, and seed damage increased markedly when SPRS exceeded 300 r·min−1. Initial seed stock in the seed chamber (ISIC) mainly affected local crowding and compressive forces, with higher ISIC increasing the risks of compression and shear damage. Based on a central composite design and multi-objective optimization, the optimized low-damage operating parameter combination was SPRS = 297.8 r·min−1 and ISIC = 37, yielding a measured seed breakage rate of 0.5 % and a germination rate of 97.75 %. Critical damage speed tests indicated a critical value of 300 r·min−1, and multi-variety tests showed variety-dependent tolerance to mechanical loading. These findings provide guidance for low-damage design and operating parameter optimization of centrifugal high-speed precision seed metering devices.

This article details a strawberry picking robot for raised-bed cultivation (with side-ripe strawberries). The robot, on a mobile platform straddling the bed, has two newly-designed picking arms. Each arm has a mechanical gripper and an RGB camera. It picks by targeting the strawberry stem, reducing damage. The specially designed arms, grippers, picking process, and sequence decision allow for unmanned autonomous operation. This reduces the strawberry picking cycle time and minimizes strawberry damage. Field test results show that the average cycle time for a single robotic arm to pick one strawberry is 5 s. For the harvester equipped with two robotic arms, the average cycle time to harvest a strawberry is 4.57 s. The success rate of damage-free picking is 59.56 %, while the overall success rate, including 'damage,' is 66.8 %. In cases where the strawberry stem is completely obscured or the strawberry is heavily covered by other ripe strawberries, the success rate of strawberry harvesting is 70.98 %. The robot shows good picking speed and success rate. Analyzing failed cases, most failures result from detection algorithm localization and end-effector sensor errors during picking.

Discharge uniformity and metering accuracy are essential for precision fertilization. An inclined screw fertilizer metering device with a spherical buffering outlet was designed and optimized using the discrete element method (DEM) and bench experiments. A particle-flow model was established using EDEM (DEM simulation software) to evaluate the effects of conveying inclination angle and outlet accumulation length on discharge performance using the coefficient of variation (CV), standard deviation, amplitude, and pulsation degree. Single-factor tests indicated that an inclination angle of 12° and an accumulation length of 32 mm produced favorable discharge performance. Two-factor optimization identified 13.5° and 40 mm as the optimal combination, yielding the lowest CV (2.36%) and standard deviation (51.88). Bench experiments further verified the performance of the optimized device. Compared with a conventional fluted-wheel fertilizer metering device, the optimized screw device reduced the average discharge error from 8.94% to 3.35% and achieved average CV values of 5.91%, 4.39%, and 4.44% under low-, medium-, and high-discharge-rate conditions, respectively. These results show that optimizing the inclination angle and outlet accumulation length can effectively improve fertilizer discharge uniformity and metering accuracy.

To achieve precise identification, length estimation, and pruning point localization for robotic tomato pruning, this study presents a method for branch length recognition and pruning point localization based on an improved YOLOv8 model. The proposed YOLOv8n-CE integrates the CBAM attention mechanism to enhance the model's focus on critical branch features, and replaces the original loss with the EIOU loss to improve bounding box regression accuracy and convergence speed. Instance segmentation is subsequently performed within the detection boxes to obtain the main stem mask and extract the upper and lower endpoints of the stem. The pixel coordinates of these endpoints are transformed into three-dimensional camera coordinates to compute the stem length, and the actual branch length is derived by combining the pixel height ratio of the bounding box with the segmentation mask. The pruning point is localized along the central line of the main stem, 1.5 cm above the lower endpoint, ensuring reduced pathogen intrusion and faster wound healing. Experimental results demonstrate that the YOLOv8n-CE model improves mAP50 by 1.7 percentage points compared with YOLOv8n. For branch length measurement, the method achieves an R2 of 0.929, an MAE of 0.411 cm, and an RMSE of 0.494 cm. The pruning point localization success rate reaches 92%, with a mean absolute error of 0.247 cm. These results verify that the proposed approach meets the accuracy requirements for tomato branch measurement and pruning point localization, providing a reliable theoretical and technical foundation for robotic pruning applications.

Early and rapid detection of herbicide-induced stress in maize is crucial for mitigating toxicity and promoting sustainable field management. Although numerous methods and equipment have been developed to monitor plant stress, many of these technologies face significant limitations, such as high costs, low accuracy, and limited integration, which hinder their practical application. In this study, we developed a portable multispectral device equipped with deep learning techniques to enable rapid, non-destructive detection of nicosulfuron-induced stress in maize. A herbicide stress experiment was first conducted on maize plants under field conditions. Spectral data and corresponding physiological indicators-SPAD value, leaf water content, and herbicide residue-were collected on days 2, 4, 7, and 10 after herbicide application. Based on these data, a multi-branch deep learning model was developed for both regression of physiological and biochemical indices and classification of stress levels. The trained model was subsequently integrated into the device, enabling on-site data acquisition, realtime computation, and visualization of results. Our results indicate that after 4 days of nicosulfuron treatment, the R2 values for SPAD, water content, and residue regression predictions were 0.79, 0.75, and 0.73, respectively, with a classification accuracy of 91.2% for stress levels. In tests using datasets from different years, the R2 values for SPAD, water content, and residue were 0.76, 0.73, and 0.70, with a classification accuracy of 87.91%. For different maize varieties, the regression R2 values were 0.71, 0.71 and 0.67, with a classification accuracy of 82.41%, demonstrating strong generalization and reliability. This study provides an effective solution for the precise detection of herbicide stress in maize fields.

Seeding depth consistency and seedling emergence uniformity significantly influence maize crop yield. To address the reduction in sowing quality caused by unreasonable changes in sowing depth, this study proposed a Gauge Wheel Load (GWL) control system. First, the factors influencing the GWL were analysed, and a corresponding mechanical model was constructed. Second, a novel measurement method was developed to accurate capture GWL by detecting the deformation on both sides of the gauge wheel arms - filling the current research gap related to seeding depth fluctuations resulting from inconsistency load on both sides. Subsequently, based on various approaches, a GWL control model was constructed, different control strategies were compared, and a comprehensive GWL control system was integrated. Finally, the proposed model and system were validated through field tests. The obtained results show that, compared to the traditional spring planter, the new GWL control model increased the qualified sowing depth rate by 8.03%∼21.53%, decreased the variation coefficient of sowing depth by 2.1%∼9.67%, and reduced the variation coefficient of emergence time by −0.59∼10.03%. These findings underscore the significant improvements in sowing depth stability and emergence uniformity. Overall, this study provides valuable technical guidance for improving sowing quality and reporting maize yield and quality.

Stable seed feeding is essential for the high-quality operation of a centrifugal high-speed maize precision metering system. To address the performance degradation caused by poor seed feeding stability, this study developed a shaped-hole wheel seed feeding device. Using the discrete element method-multi flexible body dynamics (DEM-MFBD) co-simulation, the major axis length (L) and minor axis length (W) of the shaped holes were identified as key factors influencing seed feeding stability, and their appropriate ranges were determined. Parameter optimisation yielded an optimal combination of L = 20.98 mm, W = 14.90 mm, and a working speed (V0) of 12 km h-1. Under these conditions, the absolute deviation of seed feeding rate was 0.093 seeds s-1, the coefficient of variation of seed feeding rate was 0.593%, and the range of seed feeding quantity was 6 seeds. Experiments indicated that, within the working speed range of 12-18 km h-1, the seed feeding stability of the shaped-hole wheel seed feeding device was superior to that of the groove-wheel seed feeding device. Specifically, at a working speed of 18 km h-1, the metering system equipped with the shaped-hole wheel seed feeding device achieved a qualified index of 94.7%, a multiple index of 2.8%, and a miss index of 2.5%. Compared with the system equipped with the groove-wheel seed feeding device, the qualified index increased by 1.9%, while the multiple index and miss index decreased by 1.2% and 0.7%, respectively. These results demonstrated that the shaped-hole wheel seed feeding device enhanced the performance of the centrifugal seed metering system.

To address the issue of inconsistent sowing depth caused by the forward direction surface slope (FDSS) in hilly and mountainous areas, this study developed a sowing depth control system (SDCS) based on FDSS (FDSS-SDCS), formulated a mathematical model relating FDSS, operation speed (OS), and optimal downforce, and proposed a sowing depth control strategy based on FDSS segmentation. The accuracy and response time of the shaft pin sensor and FDSS-SDCS were calibrated through indoor experiments. The experimental results demonstrate a strong correlation between the actual downforce and the response voltage output from the shaft pin sensor, with an R2 value of 0.9995; The response time of the FDSS-SDCS reaches a maximum of 0.65 s, with a maximum steady-state error of 1.70 N and a maximum overshoot of 3.63 %, all of which meet the requirements for sowing depth operations under FDSS conditions in hilly and mountainous areas. The field experiment results of the FDSS-SDCS indicate that, compared to traditional mechanical profiling springs (with spring initial increments of 10 mm, 30 mm, and 50 mm), the FDSS-SDCS with constant downforce stability control (output downforces of 300 N, 800 N, and 1300 N) reduced the maximum difference in average downforce at adjacent sampling points by 359.88 N, 435.71 N, and 467.72 N, respectively. The maximum difference in the average downforce of the FDSS-SDCS at different OSs ranged from 366.49 N to 430.75 N, indicating a significant improvement in the stability of the actual downforce under active control by the FDSS-SDCS; The range of the average sowing depth (ASD) under active control was reduced from 50 +/- 3.22 mm to 50 +/- 0.88 mm. Additionally, the qualified rate of sowing depth (QRSD) increased by 0.20 % to 17.61 %, while the coefficient of variation in sowing depth (CVSD) decreased by 0.27 % to 3.16 %. The FDSS-SDCS developed by this institute is suitable for sowing depth operations under FDSS conditions in hilly and mountainous areas, bringing the sowing depth operation performance under FDSS closer to that of non-tilting state surface.

To address the issue of low ignition rates observed with planar friction discs when testing common primary explosives in a rotary friction sensitivity tester, this study designed two friction module variants with baffled grooves (Type 1: wide groove; Type 2: narrow groove). Structural stress-strain simulations were performed using ANSYS Workbench, followed by systematic friction sensitivity tests on carbohydrazide perchlorate primary explosives (GTX, GTM, GTN). Experimental results demonstrated that the planar friction disc (Type 0) failed to initiate the explosives due to its large contact area and the propensity of the agent to spin off. The Type 1 friction disc successfully initiated GTX under conditions of 1000 N and 5 r/s (ignition time < 1 s), and also exhibited initiation capability for GTM and GTN under certain operational conditions, albeit within a relatively narrow testing range. The Type 2 friction disc exhibited superior performance, effectively initiating GTX across a broader range of conditions (e.g., 4-5 r/s at 1000 N; 5 r/s at 800 N). Furthermore, test results for GTM and GTN using the Type 2 disc demonstrated excellent regularity, with ignition time decreasing as pressure and rotational speed increased—a trend consistent with data obtained from the traditional MGY-type friction sensitivity tester. This study confirms that optimizing the friction disc structure to increase local pressure and suppress agent spin-off significantly enhances the testing effectiveness and application scope of the rotary friction sensitivity tester.

The centrifugal high-speed precision seed metering device has significant potential for improvement in seeding speed, but its seeding performance still requires further enhancement. This paper addresses the issue of uneven seed distribution within the seed metering device by designing a short hole lug structure and optimising the length of the agitator vane. The aim is to improve seed distribution uniformity and enhance seeding quality. By combining discrete element method and computational fluid dynamics in simulation experiments, the structural parameters of the short hole lug and agitator vane are optimised. The results indicate that the short hole lug structure and an appropriate agitator vane length can significantly improve the seed distribution uniformity within the seed metering device and enhance seeding quality. The 75 mm agitator vane length demonstrated the best performance in seeding efficiency. Using the response surface methodology, the optimal structural parameters for the short hole lug were determined: guide vane height of 7.94 mm, hole lug height of 4.15 mm, and hole lug length of 15.08 mm. Compared to the traditional long hole lug structure, the short hole lug structure increased the seeding pass rate by 2.3%. Field test results showed that the centrifugal high-speed precision seed metering device achieved a pass rate >= 91.6%, a miss rate <= 3.8%, and a multiple rate <= 5.4% within a seeding speed range of 12-18 km h-1. This demonstrates that the optimised centrifugal high-speed precision seed metering device is effectively reliable and suitable for higher seeding speeds.

To overcome the inefficiency and subjectivity of manual seedling surveys, this study presents a unsupervised framework for evaluating maize sowing quality and emergence uniformity via UAV-based remote sensing. Centimeter-level multispectral imagery was captured to reconstruct 3D point clouds using SfM and MVS techniques. At the algorithmic level, an improved unsupervised pipeline was developed: the Otsu method was employed for plant segmentation, followed by a Fourier Transform to extract 2D spatial frequency features for precise crop row identification and automated spacing measurement. Subsequently, the Combined Entropy Uniformity (CEU) index was developed using Shannon entropy, and a proxy for canopy closure (CCP) was derived using a porosity model, thereby enabling the simultaneous relative quantification of seedling height consistency, spatial distribution uniformity, and canopy geometric structure variability. At the application level, the framework was validated through field trials involving 19 precision planters of diverse configurations. Performance was assessed using indices such as qualified spacing, miss-sowing, and the Coefficient of Variation of Plant Spacing (PSCV). Results indicate that: (1) Vacuum-type planters exhibited optimal stability at speeds of 7–9 km/h, achieving an average qualified spacing rate of 76.7% and a PSCV of approximately 24%, whereas finger-pickup planters were more sensitive to seed size variation and mechanical vibration. (2) The results from the Generalized Additive Model (GAM) suggest a possible nonlinear relationship between seeding rate and certain uniformity indices, indicating that appropriately adjusting operational parameters could help balance operational efficiency and seeding quality; however, this trend requires further validation with larger sample sizes and repeated observations. (3) Point cloud CEU metrics and canopy structure proxies based on the Gap Fraction model showed statistical correlations with certain manually collected indicators, indicating that this method has the potential for rapid screening of seeding quality and relative evaluation of seedling population structure at the field scale under the current experimental conditions.

With the advancement of smart agriculture, precision variable-rate seeding requires high-resolution soil information. However, existing methods still fall short in generating localized soil property maps in real time, limiting inter-row seeding control for wide-span planters. To address this challenge, this paper proposes the Local Soil property distribution maps Generation Network (LSGN), based on the Wasserstein Generative Adversarial Network with Gradient Penalty and Mean squared error (WGAN-GPM). This proposed model integrates a Vision Transformer (ViT) encoder with convolutional and deconvolutional Residual Networks (ResNet), significantly improving feature extraction from discrete soil property data and enhancing the reconstruction accuracy of localized soil maps. The model adopts a two-stage training strategy. First, it performs self-supervised pre-training using local soil property maps with neighborhood information, enabling the network to learn spatial correlations and improve boundary prediction. Second, the pretrained weights are transferred into a WGAN-GPM adversarial training framework, which combines Wasserstein distance, gradient penalty, and Mean Squared Error (MSE) loss to jointly optimize the generator and discriminator. This ensures both pixel-level accuracy and distributional consistency. Compared to standard Generative Adversarial Networks (GANs), the WGAN-GPM-trained LSGN prediction reduces Relative Error Average (REA) by over 1.02 %, RMSE by over 0.16, and Kullback-Leibler Divergence (KLD) by over 1.76 x 10-4, while increasing Peak Signal-to-Noise Ratio (PSNR) by over 2.43. LSGN achieves REA values of 0.41 %, 1.19 %, and 1.80 % for planting widths of four, six, and eight rows, respectively, outperforming traditional generative models. Field cross-validation further shows that, relative to Kriging interpolation, LSGN reduces boundary region prediction errors by up to 2.42 % and central region errors by 0.97 %. Overall, the REA in local distribution maps decreases by more than 1.5 %, and RMSE is reduced by over 0.4. This study demonstrates that LSGN enables real-time high-precision soil mapping, providing a practical solution for precision seeding. (c) 2025 The Authors. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co., Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

The seed viability detection before sowing is indispensable in the agricultural production of mung beans. The conventional detection methods for seed viability are destructive, carry a risk of contamination, and fail to identify individual non-viable seeds. In this study, an efficient and sustainable method for online viability detection of mung bean seeds was developed, which utilized hyperspectral techniques and had characteristics of rapid speed, non-destructive analysis, and the ability to detect the viability status without pollution to the environment. A sample holder for mung bean seeds was designed to stably collect spectral data. The effects of different optimal spectral bands and modeling algorithms on the detection accuracy of seed viability were analyzed. Compared to the support vector machine (SVM) and the extreme learning machine (ELM) algorithms, the partial least squares (PLS) algorithm based on the visible and near-infrared spectra (380~980 nm) had better performance. The accuracy for the identification of non-viable seeds was 98.8%, and the error of viability prediction was 20.71%. The cost of a one-time viability test is $0.25 with energy consumption of 0.05 kWh−1, which is much lower than the germination test with a cost of $80.2 and energy consumption of 50.4 kWh−1. Furthermore, individual non-viable seeds can be identified and removed, and the revenue increases by $286.9 per hectare after sorting the non-viable seeds from the seeds with an 85% germination rate. This will promote the cleaner production of mung beans without additional chemical solutions added in the process.

As electric-driven maize precision planters evolve toward higher operating speeds and multi-row configurations, the power demand of the electric drive system increases. How to maintain seeding accuracy under high-speed operation while mitigating the impacts of elevated energy consumption on machine endurance and operational reliability has become a critical bottleneck limiting performance improvements in seeding equipment. To address these challenges, this study developed an Energy-efficient High-Precision Electric Seed Metering System (EHPE-SMS). The core contribution of this study is the development of a coordinated control architecture integrating Linear Active Disturbance Rejection Control and Simplified Field-Oriented Control for a two-phase (LADRC–SFOC), four-wire hybrid stepper motor, which jointly optimizes the speed loop and current loop to improve disturbance rejection and energy utilization efficiency. To support practical deployment of this control framework, a machine-learning-based fitting and optimization method was further used to tune key LADRC parameters, and an integrated drive-and-monitoring circuit was developed to provide step-loss detection, seed blockage detection, and seeding quality monitoring. Bench tests show that, under a regulated 12 V supply, the proposed EHPE-SMS has the lowest calculated electrical input power among the tested motor-drive combinations, with reductions of 70.90%, 60.29%, and 53.39% relative to the tested brushed DC motor-drive, commercial closed-loop stepper motor-drive, and brushless DC motor-drive combinations, respectively. Under identical seeding protocols on the same hardware, LADRC-SFOC further reduced average operating power by 8.04% relative to conventional PID control. Across a speed range of 3–24 km/h, relative to conventional PID control, the proposed LADRC-SFOC strategy reduces the average coefficient of variation of motor speed by 3.92% under constant-speed conditions and shortens the average response time by 406.33 ms; under both constant-speed and noise-disturbance modes, it reduces the average coefficient of variation of seed spacing by 21.7 percentage points and lowers average power consumption by 8.04%. Field experiments further demonstrate that, over the same speed range, the system achieves an average seeding quality index (QI) of 95.04% and an average seed-spacing coefficient of variation (CV) of 24.54%. However, due to factors such as seed bounce under high-speed conditions, seed-spacing uniformity remained somewhat limited at speeds above 15 km/h. Moreover, the integrated monitoring module successfully detected and responded to the tested abnormal events, indicating its functional effectiveness in improving system safety. Overall, the proposed EHPE-SMS substantially reduces energy consumption while maintaining high-speed, high-precision seeding performance, and enhances system stability and intelligence, supporting efficient, clean, and sustainable precision seeding equipment.

Optimal seeding depth (SD) is crucial for ensuring uniform and robust maize seedlings, significantly impacting yield. However, traditional manual adjustment methods for seeding depth are labor-intensive and cannot adapt dynamically to varying soil conditions. This study developed and evaluated an electric-driven depth control system for maize planters that allows for precise, automated, and rapid retrofitting. The system incorporates a self-locking worm gear mechanism to ensure stable depth adjustment and features a quick-installation base plate that enables convenient upgrades without altering the planter's original construction. A proportional-integral-derivative (PID) closed-loop control algorithm based on motor rotation counts was established, achieving a coefficient of determination (R2) of 0.994 in the depth calibration model, with a maximum overshoot of 3.533% and steady-state error below 5% in control response tests. Soil bin experiments demonstrated that under operation speeds (OS) of 1-9 km/h and seeding depths of 20-80 mm, the system achieved high accuracy, with the average seeding depth (ASD) deviating from the target by a maximum of 3.16 mm. The system maintained a qualification rate of seeding depth (QRSD) of at least 88.01% (average 96.29%), with a coefficient of variation of seeding depth (CVSD) averaging 7.29%. The system also exhibited high stability, with QRSD for deep and shallow adjustments of 99.04% and 97.14%, respectively. Compared to manual adjustment, the electric-driven system offers the significant advantages of stepless control and rapid adjustment. The results demonstrate that the system provides a viable technological foundation for dynamic seeding depth adjustment in response to diverse soil conditions, contributing to improved agricultural productivity.

Accurate, nondestructive assessment of fresh tea leaf quality is important for breeding and field management, yet most spectral work still targets processed or low-moisture products. Here, a mechanistically guided hyperspectral method was developed to estimate free amino acids (AA) and total polyphenols (TP) in fresh leaves. Spectral experiments on purified AA and TP powders and their water mixtures identified a key spectral window at 1660 nm. Fractional-order derivatives were applied to leaf reflectance spectra from 102 spring samples (53 varieties), and full-spectrum Partial Least Squares Regression (PLSR) models were used as comparison and validated on an independent set of 40 summer samples. PLSR achieved decent cross-validation coefficient of determination accuracy for AA (Rcv2=0.867) and TP (Rcv2=0.755) and good external prediction coefficient of determination accuracy (RP2=0.793 and 0.776, respectively). Guided by the powder and leaf-level analysis, two-band NDSI indices were derived: the AA index of 1735/1626 nm (R2ₚ = 0.687, RPDₚ = 1.788) and the TP index of 1673/1660 nm (R2ₚ = 0.785, RPDₚ = 2.157) approached the PLSR, indicating that much of the useful information for AA and TP is concentrated in this narrow window and can be captured by simple, interpretable indices potentially suitable for in-field sensing, pending validation across multiple sites, seasons, and management conditions.