Ratoon rice, defined as the production of a second crop from the stubbles remaining after harvest of the main crop, was increasingly promoted in China as a strategy to enhance annual grain yield on limited arable land without additional land preparation, sowing, or transplanting. With the rapid adoption of combine harvesters in major rice-growing regions, mechanized rice ratooning technology with the main crop harvested mechanically became the predominant method in ratoon rice. However, mechanical harvesting introduced critical constraints to ratoon crop. Track-induced crushing, excessive ground pressure, and unstable cutting height frequently caused stubble breakage and axillary bud injury, thereby reducing bud survival, suppressing ratoon tiller production, and compromising yield formation and milling quality in the ratoon crop. These mechanical damages emerged as a primary bottleneck limiting the stable and large-scale adoption of mechanized rice ratooning system. From an agricultural engineering perspective, this review systematically synthesized recent advances in understanding damage mechanisms and in developing technical strategies for loss mitigation, yield stabilization, and quality improvement in mechanized rice ratooning system, with emphasis on machinery optimization, varietal improvement, and integrated agronomic management. Mechanistic studies demonstrated that ground contact pressure and shear forces generated by harvester tracks disrupted stubble structural integrity, damaged vascular tissues, and impeded assimilate translocation to regenerated buds. Under high soil moisture conditions, increased sinkage and soil deformation amplified mechanical stress, substantially reducing bud sprouting rates within track zones compared with non-track areas. In crushed zones, ratoon development often shifted from upper-node buds to lower-node buds with delayed phenology, resulting in reduced canopy uniformity, decreased effective panicle number, and significant yield penalties. Delayed panicle emergence in track areas also led to asynchronous maturity within fields, thereby decreasing head rice rate and increasing variability in milling quality. To quantify mechanical damage, studies adopted indicators including missing-stubble rate, bud survival rate, ratoon tiller-to-panicle conversion rate, yield loss rate, head rice rate, and chalkiness-related parameters. Emerging technologies such as unmanned aerial vehicle remote sensing, machine vision, and in-field sensors were increasingly applied to identify track zones and characterize spatial heterogeneity of mechanical damage, although standardized evaluation protocols remain insufficient. Engineering innovations primarily targeted reductions in crushed area and crushing intensity, as well as improvements in stubble-height uniformity. Advances included lightweight chassis designs with reduced ground pressure, optimized track width and cutting width configurations, automatic header-height control based on multi-sensor perception systems, stubble-righting devices integrated with harvesters, and navigation-assisted path planning to minimize track overlap and headland damage. Although these technologies effectively mitigated mechanical impact and improved ratoon crop performance under experimental conditions, trade-offs among ground pressure, machine stability, fuel consumption, operational cost, and field adaptability limited widespread commercial application. Varietal differences in stem mechanical strength, ratooning ability, and non-structural carbohydrate reserves significantly influenced tolerance to mechanical harvesting stress. Evaluation metrics extended beyond ratoon yield to include bud survival rate, ratoon panicle number, stem morphological traits, and biomechanical properties. Although quantitative trait loci and candidate genes associated with ratooning ability and stem strength were reported, stable loci and deployable molecular markers specifically targeting crushing tolerance remain limited. Complementary agronomic practices, including skip-row planting, pre-harvest drainage to enhance soil bearing capacity, timely post-harvest nitrogen topdressing with balanced phosphorus and potassium inputs, and targeted rehabilitation of track zones, partially alleviated yield and quality losses. Overall, the accumulated evidence indicated that future progress in mechanized rice ratooning system depended on coordinated machine, variety, and agronomy integration, lightweight and intelligent harvester development, standardized damage evaluation systems, and digital monitoring platforms to ensure stable, scalable, and quality-oriented ratoon rice production.

Conventional Miner’s linear cumulative damage can focus only on the stress amplitude. In this study, a dynamic damage assessment was proposed for the high-horsepower tractor using the measured load spectrum. The research object was taken as the transmission shaft of a 162 kW four-wheel drive wheeled tractor. Multidimensional parameters were integrated to employ a dynamic weight allocation. A framework of the damage assessment was also established using the loads in the time domain, energy distribution in the frequency domain, and fluctuation features. Firstly, a torque test was developed using the transmission shaft of a tractor. The high-precision strain torque sensors were integrated with the wireless data acquisition. Torque fluctuation signals were captured under various conditions in real time, such as field operations and road transportation. The wheel torque signals were measured in the field, and then converted into the equivalent torque at the transmission shaft, according to the wheel-side planetary reduction mechanism, central transmission, transfer case, creeper gear, and gearbox. Subsequently, the rainflow counting was applied to identify the load cycles in the torque data. Statistical features of the loads were derived from high-confidence data for subsequent analysis. A composite model of the "three-dimensional linear weighting–nonlinear compression" was proposed to construct the damage indicator. Furthermore, three indicators—load gradient, frequency-band damage energy, and load kurtosis—were linearly weighted. The weighting parameters of these three indicators were optimized using a grid search algorithm. The root mean square error and absolute percentage error were employed to quantitatively evaluate the performance of different parameter combinations. The optimal weight allocation was determined as 0.350, 0.425, and 0.225. An S-shaped compression mapping with the sigmoid function was introduced to mitigate the influence of extreme values on the assessment, effectively balancing the sensitivity and robustness. The center of the load cycle time was indexed within the intensity index window of the dynamic damage. Time-frequency features were integrated into the cyclic damage for the amplitude correction. A hazardous threshold with the dynamic damage intensity index represented the exponential amplification of the cyclic amplitudes. The Goodman formula was then used to convert the torque signals into equivalent zero-mean stress amplitudes. And the S-N curve of the transmission shaft was applied to compute the damage. As such, a systematic assessment was realized on the dynamic damage. Experimental results demonstrate that the improved model effectively calculated the damage over the load bands. The peak normalized power spectral density energy occurred at 0.1 Hz, while the maximum damage frequency band was at 0.43 Hz, with a damage value of 1.78×10^-8. There was no coincidence between the peak power spectral density energy and the maximum damage. The theoretical correctness of the improved model was validated because the conventional approach only considered the amplitude. The relative error of the conventional Miner’s damage rule was 51.11%, whereas the relative error of the damage prediction was only 0.67%. The high accuracy and feasibility exhibited in the damage prediction. The linear damage models were advanced beyond the loading sequence and statistical analysis, thus providing the theoretical basis and insights for future assessment of the damage. Furthermore, a three-dimensional space of the damage feature was incorporated with the load gradient, frequency-band energy, and load kurtosis. The multidimensional parameter mechanism can be directly applied to the life prediction of the rotating machinery, such as the transmission shafts and gearboxes, or the damage assessment in the structural domains, such as the composite materials and additive manufacturing components. The finding can provide a theoretical and quantitative tool for the intelligent operation and prediction.

Ecological ditch–pond systems are important measures for controlling agricultural non-point source pollution, yet their practical application is constrained by unstable purification performance and large land occupancy. The effective application of such systems in irrigation districts depends not only on the design and operation management of individual units, but also significantly on their spatial layout (including system area and unit connection pattern). Existing studies often fail to adequately capture the multi-level dynamic responses of water volume and water quality in such systems, which hinders their support for spatial layout optimization. Against this backdrop, this study proposes a system dynamics-based simulation method for optimizing the spatial layout of ditch-pond systems in irrigation districts. A field-ditch-pond system model was developed using the system dynamics simulation tool Vensim, integrating water balance, pollutant removal processes, and hydraulic connections among ditches and ponds. The water depth in the paddy model was determined by inflows, outflows, and water consumption during each time step, while the total nitrogen and total phosphorus concentrations were simulated by considering fertilization, first-order pollutant decay, and inputs from rainfall and irrigation. For the ditch–pond unit model, water volume changes were governed by rainfall, evapotranspiration, seepage, upstream inflow, and drainage discharge. Pollutant concentrations in the ditch–pond unit model were calculated using two modes: static storage-based reduction and dynamic drainage-based reduction. The paddy and ditch–pond unit models were linked through system dynamics into an integrated field-ditch-pond system model, which was calibrated and validated using field monitoring data. A case study was conducted in a typical double-cropping paddy high-standard farmland demonstration area in southern China. The model verification results show that the developed model can effectively simulate the dynamic variations of water volume and pollutant concentrations in the system. The case analysis results show that: 1) With increasing ditch-pond to paddy area ratio, nitrogen and phosphorus removal rates rise, but the rate of increase gradually slows down, suggesting an optimal range of 5%-9%; 2) During the late rice season or under larger area ratios, concentrating wetlands in a single drainage path results in a significantly lower removal rate compared to other layouts; 3) For practical implementation, it is recommended to first determine an appropriate area ratio based on target pollutant reduction goals. Subsequently, wetlands may be placed either at the main drainage outlet or distributed in parallel across different drainage pathways, depending on site-specific conditions. The findings provide a methodological reference for modeling multi-level wetland systems and offer a scientific basis for ecological control of agricultural non-point source pollution. Future work may further refine the simulation of water cycling and pollutant transformation processes within field-ditch-pond systems to enhance model accuracy and applicability.

Robotic arms are often required for high picking efficiency, path planning, and path smoothness for safflower harvesting in unstructured environments. This study aims to introduce a picking area clustering and a Target redirecting rapidly-exploring random tree (TR-RRT*) path planning. Firstly, a roller end-effector was designed with an effective picking area of 5 cm × 20 cm, according to the spatial distribution of safflower seed balls under natural conditions. A picking point clustering was also developed to divide the working range of the robotic arm into multiple sub-areas. Each cluster was designed to cover 1-3 picking points. The end-effector was used to harvest 1-3 safflowers in a single operation. A goal redirecting, a goal-direct and deflection expansion, and an artificial potential field (APF) tangential escape force strategy were integrated into the bidirectional RRT* framework to improve the path search efficiency and obstacle avoidance. Among them, the goal redirecting strategy continuously updated the target points during expansion to rapidly connect paths within opportunity windows. The goal-directed and deflection expansion allowed both search trees to extend from the nodes closest to the goal, thus maximizing progress toward the target while deflecting to avoid obstacles. The tangential escape force strategy introduced a tangential component into the conventional artificial potential field, enabling smooth sliding along obstacle boundaries when approaching them. As such, the module effectively avoided the path oscillation and target unreachability in the conventional APF. In path optimization, a combination of a greedy jump-point strategy, interpolated curvature optimization, and B-spline curve fitting was applied to smoothly adjust the curvature of path nodes and then generate high-order continuous trajectories. An obstacle avoidance reconstruction was also introduced to prevent the trajectory penetration through obstacles for the continuous, collision-free, and smooth motion of the robotic arm. A comparison was made on the TR-RRT* and seven algorithms in the dense, small-obstacle environments. The results demonstrated that the superior performance was achieved in the complex obstacle environments, where the path lengths (3892.05mm) were shortened by 9.16% and 8.22%, compared with the IBI-P-RRT*(4284.44 mm) and BI-RRT*(4240.45 mm), respectively; While the average planning time (0.30 s) was only 7.94% of that of RRT*(3.78 s) and 69.77% of BI-RRT*(0.43 s), respectively. Additionally, the steering angles of BI-RRT*(37.51°) and BI-APF-RRT*(39.61°) were 1.22 and 1.29 times larger than those of the improved algorithm (30.63°), indicating significant improvement of the path smoothness. Statistical quantitative experiments were conducted for the algorithm in different obstacle environments. The optimal performance was achieved in both path length and time consumption under various environments. Moreover, the TR-RRT* algorithm successfully planned paths in narrow passage obstacle environments. The tendency of conventional APF was to avoid the local oscillations during path planning in such scenarios. Physical harvesting tests further validated the effectiveness of the clustering and TR-RRT* algorithm. The robotic arm took an average of 3.64 s to move from the initial position to the first target point, with an average transfer time of 3.12 s between tasks. The average positional error relative to the robotic arm's workspace was less than 0.9%, indicating the stable and efficient performance of safflower harvesting.

High-load disturbances during rotary tillage can cause significant wheel slip on the electric-drive mobile platform in the distributed horticulture facility. It is often required to control the speed-slip rate for the longitudinal stability of the distributed horticultural facility. In this study, a cascaded controller of vehicle speed–slip rate was proposed using integral robust vehicle speed and sliding-mode slip rate control. Its effectiveness was validated using simulations and vehicle experiments. Firstly, a dynamic model was established for the coupled system between the distributed electric-drive horticultural platform and the rotary tiller. Tire-soil interaction and the resistance of rotary tillage were also considered to explicitly incorporate the wheel rotational dynamics and external disturbance torques. Soil adhesion also led to variable tillage resistance. Moreover, a coupled modeling framework was constructed to describe the nonlinear relationship between longitudinal tire force and slip ratio. The traction generation was accurately characterized under deformable soil conditions. The slip regulation and speed stabilization were coordinated under high-load environments. Subsequently, an outer-loop vehicle speed controller was designed to incorporate integral robust control. The steady-state errors were eliminated from the operational disturbances for the high-speed stability. The integral term was used to compensate for the persistent disturbance-induced bias. While the robust component was enhanced, the controller’s tolerance to parametric uncertainties and unmodeled dynamics. Integral action was combined with robustness enhancement. The outer-loop controller maintained accurate speed tracking, even when sudden load fluctuations occurred. Furthermore, an inner-loop slip rate controller was developed using sliding-mode control. The optimal slip rate was obtained from the inverse tire model to serve as the reference input for the rapid convergence and precise tracking of slip rate. Sliding-mode control was selected for its high robustness against disturbances and modeling uncertainties, thereby enabling the dynamic response and strong anti-interference. The optimal slip rate corresponded to the traction peak region of the tire–soil interaction curve. Traction efficiency was maximized to prevent excessive slip. The outer and inner loops were coordinated for the longitudinal stability of the platform under high disturbance. A control strategy was then integrated for anti-slip driving and speed regulation. Specifically, the inner loop was used to rapidly suppress the deviations of the slip ratio, while the outer loop was for the global speed regulation using a cascaded structure. A hierarchical architecture of traction control was constructed, suitable for the distributed electric-drive systems. Simulation results indicate that the cascaded controller achieved an average speed error of 0.10 km/h under sudden muddy conditions, which was reduced by 16.6% and 67.7%, compared with the switching and speed control, respectively. The speed recovery time was 0.11 s, which was reduced by 64.5% and 68.6%, respectively. There was an average speed error of 0.07 km/h under variable tillage depths, which was reduced by 40.0% and 52.0%, compared with switching control and speed control, respectively. Experimental results indicate that an average speed error of 0.44 km/h was found under acceleration, which was reduced by 4.3% and 8.3%, compared with the switching and speed control, respectively. The average slip ratio was 0.15, which was reduced by 11.7% and 16.6%, respectively. There was an average speed error of 0.20 km/h under deep tillage, which was reduced by 20.0% and 37.5%, respectively. The average slip ratio was 0.13, which was reduced by 23.5% and 31.6%, respectively. Therefore, the cascaded controller can be expected to effectively suppress the slip ratio during rotary tillage, thereby enhancing the speed control performance and operational stability.

Thermal and humidity environments can dominate the pig growth, health status, and production performance in pig houses, including air temperature, relative humidity, and airflow velocity. The environment can be regulated to consider the interaction mechanism between housing conditions and pig thermal responses. A mechanistic and physiologically interpretable model is required to accurately simulate pig thermal responses under different thermal and humidity conditions. However, existing models of pig thermal response cannot fully meet the requirements of the intelligent control applications. In this study, a pig two-node heat exchange model (PTHM) was established using biological heat balance theory and thermodynamics. Heat exchange was also simulated among the core, the skin layer, and the surrounding environment. Metabolic heat was generated in the core layer and then transferred to the skin via tissue conduction and blood circulation. Part of the heat was dissipated to the environment as sensible respiratory heat loss. The remaining heat was stored within the body, leading to an increase in rectal temperature. Heat in the skin layer was transferred from the core via conductive transfer and blood-mediated convective transport. The heat was then dissipated to the surrounding environment via convective heat exchange and thermal radiation. A small fraction of heat was dissipated after skin evaporation. Environmental parameters were used as the model inputs, while the major physiological parameters were used as the outputs after simulations. A recognition framework of pig drinking behavior was developed using an improved YOLOv11 object detection architecture, particularly for the prediction accuracy and physiological interpretability of the model. A pig drinking detection model (PDDM) was further established to calculate drinking frequency using this framework. The drinking frequency was then introduced into the PTHM as a behavioral correction factor to regulate blood-mediated convective heat transfer and respiratory heat dissipation, thereby constructing a drinking behavior–corrected pig two-node heat exchange model (D-PTHM). A more realistic representation was obtained for the pig thermoregulation. The results showed that the air temperature was the dominant environmental factor on pig thermal physiological responses. The PTHM model also achieved coefficients of determination (R²) of 0.673, 0.685, and 0.615 for rectal temperature, heart rate, and respiratory rate, respectively. The mean absolute errors (MAE) were 0.320 °C, 7.020 bpm, and 0.916 bpm, while the root mean square errors (RMSE) were 0.412 °C, 9.120 bpm, and 1.635 bpm, respectively. A preliminary representation was obtained for the heat transfer pathway from the body core to the skin. Subsequently, the surrounding environment was offered a simplified representation of whole-body heat balance. The DCB-YOLO drinking detection model achieved a mean average precision (mAP) of 97.47%. The PDDM was used to reliably quantify the pig drinking frequency for behavioral correction of the heat exchange model. The prediction accuracy of D-PTHM was significantly improved after drinking behavior was introduced as a correction factor. The D-PTHM achieved higher R² values of 0.831, 0.771, and 0.775 for the rectal temperature, heart rate, and respiratory rate, respectively. The MAEs were 0.247 °C, 3.358 bpm, and 0.580 bpm, while the RMSEs were 0.332 °C, 4.053 bpm, and 0.747 bpm, indicating the improved model stability and environmental adaptability. The drinking behavior significantly enhanced the mechanistic model to regulate the pig thermal field under different thermal and humidity conditions. This finding can provide a physiologically realistic model for precision environmental control in pig houses. More accurate environmental regulation can be used to improve animal welfare using pig physiological responses in sustainable and efficient livestock production.

Asparagus harvesting can be confined to the efficacy of robotic vision in recent years. Asparagus spears are characterized by a slender morphology in their natural growth state. These tender stems are highly prone to mutual occlusion and overlapping when growing densely in field conditions. Furthermore, the stout mother stems can simultaneously present as the complex background interference. Collectively, it is often required for the high accuracy of the multi-target segmentation and recognition using machine vision. In this study, the lightweight instance segmentation model (YOLO11n-seg) was adopted as a baseline, in order to improve the precise positioning and harvesting performance of the robotic end-effector. Consequently, an optimized model named YOLO11n-SAL was also proposed to specifically tailor the slender, occluded targets with high fidelity. Two modules were introduced to enhance the feature extraction and attention mechanisms in the architectural framework. Firstly, the multi-scale edge enhancement Module (MEEM) was conceptually designed and integrated in order to mitigate the challenge wherein the edge features of the slender asparagus targets were inherently weak and easily lost during convolutional operations. Multi-scale decomposition was performed on the convolutional feature maps. The MEEM effectively extracted and intensified the edge and contour information before feature fusion. The sensitivity to the target boundaries was significantly elevated for the high segmentation precision, thereby enhancing the perceptual capability of the targets with the slender morphological structures. Secondly, the separated and enhancement attention module (SEAM) was introduced to rectify the feature confusion and data incompleteness caused by inter-target occlusion. Attention separation over both channel and spatial dimensions was also utilized to adaptively perceive the local and global features of the occluded asparagus at the varying scales. These features were selectively enhanced and effectively fused to better position the visible subjects of the partially masked targets, while suppressing the background noise and distractor information. The robust performance of the detection and recognition was maintained even within the complex and cluttered environments. A series of experiments was conducted to verify the effectiveness of the improved model. Quantitative evaluation results indicate that the improved YOLO11n-SAL model achieved significant gains over all key performance indicators, compared with the baseline model. In the detection task of the target bounding box, the superior performance was achieved with a detection precision of 94.2%, a recall rate of 83.1%, a mean average precision at IoU threshold 0.5 (mAP_0.5) of 91.2%, and a mean average precision at IoU threshold 0.5-0.95(mAP_0.5-0.95) of 76.2%. In the more granular instance mask segmentation, the model also performed impressively. The segmentation precision, recall, mAP_0.5 and mAP_0.5-0.95 reached 93.4%, 77.9%, 90.7%, and 62.7%, respectively. Furthermore, the heatmap analysis demonstrated that the YOLO11n-SAL model was markedly improved to perceive the asparagus edge features over different scenarios, with the superior multi-target segmentation and recognition under occluded conditions. The high accuracy of the segmentation and recognition was achieved to reduce the interference in the complex multi-scenario environments, compared with the baseline. Finally, a series of asparagus recognition, positioning, harvesting, and grasping trials were carried out using depth cameras and mechanical arms, in order to validate the cognition and position performance in the actual deployment scenarios. The empirical results showed that a positioning success rate of not less than 90% was accompanied by effective harvesting and grasping performance. These findings can provide reliable technical support for the advancement of robotic harvesting in precision agriculture.

Cropland water–land balance relationship can be regulated to maintain the cropland productivity, ecological stability, and sustainable land use. It is often required to optimize cropland spatial layout and crop planting structure in sustainable water and land resources, according to “Determining Land Use by Water Availability”. In this study, an integrated analytical framework was developed to explore the spatiotemporal characteristics of the balance between water and cultivated land resources. There was also a close relationship between crop water requirement, water deficit and surplus, drought events, and productivity loss. The research area was taken as the five major agricultural regions in northern China. Meteorological, crop planting structure, and gross primary productivity (GPP) were monthly collected from 2001 to 2022. The full-growing-season crop water requirements of winter wheat, spring maize, and summer maize were estimated using the FAO-56 crop coefficient. A decadal Crop Water Deficit and Surplus Index (CWDI) and its standardized form, the Standardized Crop Water Deficit and Surplus Index (SCWDI), were then constructed to integrate run theory. Empirical orthogonal function (EOF) analysis, correlation analysis, and a Copula–Bayesian conditional probability model were used to systematically identify the Spatiotemporal evolution of cropland water–land balance relationships, crop critical water-demand periods, and productivity loss risks under drought stress. The results showed that (1) multi-year mean full-growing-season water requirements of winter wheat and maize were 500 and 594 mm, respectively, particularly with 627 and 536 mm for spring and summer maize, respectively. Crop water requirements shared significant regional differences. In winter wheat, the water requirement followed the descending order of Huang–Huai–Hai Plain (521 mm) > Loess Plateau (514 mm) > Gansu–Xinjiang Region (467 mm), while spring maize shared the highest water requirement in the Gansu–Xinjiang Region (662 mm). (2) Average drought-event frequency ranged from 0.4 to 2.0 events per year. The Gansu–Xinjiang Region was characterized by a high proportion of extreme drought events, long duration, and high intensity, indicating a cumulative drought pattern, whereas the Huang–Huai–Hai Plain was dominated by high-frequency but short-duration drought events. (3) EOF decomposition showed that the first five modes cumulatively explained 73.7% of the total variance, with EOF1 and EOF2 accounting for 28.6% and 20.0%, respectively, indicating region-wide consistency and regional heterogeneity in interannual dry–wet variations of cropland. (4) Correlation analysis between SCWDI and standardized GPP (SGPP) showed that the key months were identified as July for spring and summer maize, while May for winter wheat. According to the cumulative effect, the water-demand stages of spring maize, summer maize, and winter wheat corresponded to May–June, June, and March–April, respectively. (5) SCWDI–SGPP dependence structures of spring maize and summer maize were best fitted by the Gaussian Copula, while the Frank Copula performed best for winter wheat. The average probabilities of productivity loss were 63%, 80%, and 44%, respectively, for spring maize, summer maize, and winter wheat under extreme drought conditions. The north-central Huang–Huai–Hai Plain and the Loess Plateau were identified as the drought stress conversion into productivity loss. The finding can provide a scientific basis for the optimal cropland layout and crop structure during drought risk prevention in northern China.

Wheel-type rice transplanters have been widely used in mechanized paddy-field transplanting because of their flexible operation and convenient field transfer. However, during paddy-field operation, the unstable adhesion state between the driving wheels and saturated paddy soil may induce lateral deviation, longitudinal slip, and row-spacing errors. To clarify the slip behavior of a wheel-type rice transplanter under different soil moisture contents, traveling speeds, and whole-machine masses, this study analyzed the variation characteristics of the slip ratios of a wheel-type rice transplanter. A Xinyang 2ZG-6D1(G4) wheel-type rice transplanter was selected as the research object. A whole-machine multi-body dynamics model was established in RecurDyn, and a layered paddy-soil particle bed was constructed in EDEM. The Hertz-Mindlin with JKR contact model was adopted as the soil contact model to describe the adhesive contact behavior among paddy-soil particles and between soil particles and soil-engaging components. Compression tests and direct shear tests were conducted to calibrate the discrete element parameters of paddy soil, thereby improving the reliability of the soil model. The bidirectional coupling between the transplanter model and the soil model was realized through the EDEM-RecurDyn coupling interface. Soil moisture content, traveling speed, and whole-machine mass were selected as experimental factors, while lateral slip ratio and longitudinal slip ratio were used as evaluation indexes. Based on the calibrated discrete element method and multi-body dynamics (DEM-MBD) coupling model, a Box-Behnken response surface experiment was conducted to investigate the effects and interaction mechanisms of the three factors on the two slip indexes. Regression models of the lateral and longitudinal slip ratios were established, and multi-objective optimization was carried out to obtain a suitable parameter combination. Field tests were finally performed to verify the prediction accuracy of the coupled simulation model. The results showed that soil moisture content had the most significant effect on both lateral and longitudinal slip ratios, with contribution rates of 20.54% and 35.74%, respectively. The interaction between soil moisture content and whole-machine mass also had an obvious influence on the two slip ratios, indicating that the effect of machine load on wheel-soil interaction depended strongly on the moisture state of paddy soil. With increasing soil moisture content, the bearing, shear, and adhesion characteristics of the soil changed, further affecting wheel sinkage, soil adhesion, and driving stability. Within the experimental range, when the soil moisture content was 32%, the traveling speed was 0.54 m/s, and the whole-machine mass was 854 kg, the lateral and longitudinal slip ratios showed relatively good comprehensive performance; that is, both indexes remained at relatively low levels under the multi-objective optimization constraints. The field validation results showed that the measured lateral slip ratio was 1.55%, while the simulated value was 1.36%, with a relative error of 12.25%. The measured longitudinal slip ratio was 11.48%, while the simulated value was 10.15%, with a relative error of 11.59%. Both errors were within 15%, indicating that the established DEM-MBD coupling model can reasonably predict the slip ratios of a wheel-type rice transplanter under paddy-field conditions. This study provides a feasible simulation method for analyzing wheel-soil interaction in paddy fields and offers a reference for the design of walking systems and the optimization of operating parameters for paddy-field machinery.

Immunization injection is of ten require d to prevent and control the diseases in pig farming. However, current needle injection still relies on manual labor in pig farms, leading to serious challenges, such as high risk of needle breakage, severe cross-infection, low efficiency, time-consuming, and labor-intensive operation, as well as missed or incorrect injections. In this study, a control system was proposed for vision-based positioning and motion in a needle-free injection robot under confined stall breeding scenarios. According to the workflow of swine immunization, the key components were selected to integrate the navigation system of the robot. Its parameters were then determined to improve the needle-free injection module. The motion range of the robotic arm was simulated using MATLAB. The rotation angles of its six axes were constrained to prevent collision between the robotic arm and confined stall. A precise localization was proposed for the needle-free injection site on pigs. The YOLOv8n framework was improved to solve the missed and false detections caused by blurred features of the tail root and susceptibility to dirt interference. The improved network was used to extract the tail root region from pig images. A point set of the hip muscle injection area was constructed for vertical injection. The least squares method (LSM) was applied to fit the surface point set, enabling accurate calculation of the injection point and posture. A control system was developed to accurately track the injection point in the needle-free injection robot, due to the random movement of pigs during operation. According to feeding system, the injection was divided into two phases: a pre-injection and an injection phase. In the pre-injection phase, the tail root position was continuously detected to perform visual servoing control of robotic arm, thereby tracking the injection point in real time. In the injection phase, the robot moved into the injection point, and then performed the injection, according to the last detected pose. The critical distance of phase transition was determined to be 15 cm using depth camera and triangulation. The robot operating system (ROS) was used to integrate navigation, visual detection, and robotic arm control algorithms, enabling the robot to follow a predetermined trajectory and then perform needle-free injection on pigs. Experiments were conducted at a pig farm in Qinhuangdao City, Hebei Province, China. The experimental results showed that the tail root detection algorithm achieved an accuracy of 95.8%, a recall of 93.5%, and a mean average precision (mAP) of 97.1%. The overall injection success rate of the robot was 93.3%, of which 95.2% were vertical injections. The maximum injection deviations in the X, Y, and Z axes were 2.9, 3.7, and 1.9 cm, respectively, corresponding to average deviations of 1.30, 1.91, and 0.59 cm. The injection accuracy was fully met the requirements of swine immunization. A closed-loop control system was realized from chassis navigation, visual recognition, dynamic tracking to precise triggering for the needle-free injection in real scenarios. The findings can also offer the technical and engineering reference for the large-scale intelligent farming.