With the increasing scale of high-clearance sprayers, large-span flexible booms are prone to severe large-deformation nonlinear vibrations under complex vertical excitations encountered in field operation, thereby markedly compromising spray uniformity and structural safety. This study aims to elucidate the global nonlinear dynamics and chaotic evolution mechanisms of flexible spray booms under forced excitation. Based on Euler-Bernoulli beam theory and the von Kármán large-deformation assumption, and taking into account the stepped cross-sectional characteristics of the boom, the structure was equivalently modeled as a two-segment stepped cantilever beam composed of an inner boom and an outer boom. The nonlinear partial differential governing equation for the forced vibration of the system was then derived using Hamilton’s principle. The natural frequencies and mode shapes of the system were obtained analytically and compared with the results of three-dimensional finite element modal simulations in ANSYS, thereby validating the accuracy of the equivalent mechanical model. Considering the low-frequency-dominated vibration characteristics observed in field conditions, the partial differential equation was further reduced, via a single-mode Galerkin truncation, to a single-degree-of-freedom Duffing-type ordinary differential equation containing geometric nonlinear terms. The harmonic balance method was employed to derive the primary resonance frequency-response and force-response characteristics of the system, revealing hardening behavior, resonance-frequency shift, and multistable jump phenomena induced by large deformation. In addition, the sensitivities of the instability boundaries to damping and nonlinear stiffness were quantitatively characterized. Furthermore, the topological route to chaos under extreme operating conditions was systematically investigated using global bifurcation diagrams, the maximum Lyapunov exponent, time histories, two-dimensional phase portraits, and Poincaré maps. The results show that the chaotic response of large flexible spray booms is highly sensitive to the amplitude of external forcing and the damping coefficient. This study provides a solid theoretical basis for vibration-reduction-oriented structural optimization and nonlinear instability control of spray booms in modern high-clearance sprayers.

Tea is one of the three major beverages in the world, and China, as the world's largest tea producer and consumer, faces challenges in addressing the poor stability and low efficiency of self-propelled tea canopy working machines when operating in mountainous tea plantations. The static leveling systems alone are insufficient to solve this issue. This paper presents a dynamic and static joint leveling control system based on Proportional-Integral-Derivative (PID), utilizing a self-developed vehicle-mounted omnidirectional leveling platform. The analysis of the platform's leveling principle and the construction of its hardware control system are discussed, then a joint control method for dynamic and static leveling is formulated. Simulation tests were conducted to determine the control parameters of the leveling system, validating the effectiveness of the control system. A prototype machine was built to test both its leveling performance and field capabilities. Results demonstrate that on a 15.0 degrees slope, the platform can reduce inclination to less than 5.0 degrees within 1.0 s, with an inclination error controlled at 0.1 degrees-0.4 degrees within 1.8-2.3 s; during field tests, mean inclination values when entering and exiting tea plantations are measured at 3.2 degrees +/- 1.5 degrees and 2.5 degrees +/- 1.4 degrees, respectively. The results indicate that this developed dynamic and static joint leveling control system enables rapid platform stabilization while reducing tipping risks in mountainous tea plantations.

Optical flow estimation is fundamental to video understanding, object tracking, and robotic perception. Although recent deep models achieve high accuracy, lightweight designs often suffer clear performance degradation in complex motion regions when feature capacity and refinement budgets are heavily constrained. Errors are particularly pronounced around motion boundaries, occlusions, textureless regions, and locally ambiguous correspondences.,To address these challenges, we propose FAL-Flow, a frequency-aware lightweight optical flow framework. Rather than treating performance loss as a simple capacity issue, FAL-Flow targets three key stages of the optical flow pipeline. It introduces a wavelet-based frequency feature compensation module for preserving structural and boundary information, a frequency-prior flow initialization module for reliable low-iteration refinement, and a Cross-Axis Motion Propagation (CAMP) module for improving motion consistency in challenging regions.,Experiments on Sintel and KITTI show that FAL-Flow improves complex-scene optical flow accuracy while maintaining lightweight efficiency. Under comparable latency to low-iteration RAFT-small, FAL-Flow reduces End-Point Error (EPE) by 21.1% and 19.9% on Sintel Clean and Final, respectively, and achieves a 29.3% relative reduction over the lightweight baseline in complex scenarios.

Wearable sensing technology offers a natural and convenient means of human-computer interaction, particularly for gesture recognition, yet domain shifts in wrist-worn single-site sensing pose significant challenges for cross-domain gesture recognition. To address this, we proposed a generalized cross-domain framework for fine-grained gesture recognition using wrist-worn single-site sensing. Concretely, we presented a Multi-Branch Network, which combines feature-level multimodal fusion with enhanced inter-modal interaction to effectively capture fine-grained gestures. To this end, we constructed a multimodal dataset, which comprises fifteen static and eighteen dynamic gestures. Furthermore, we developed five fine-tuning strategies and evaluated them across the paradigms of cross-session, cross-subject, cross-gesture, and cross-modality. Through comprehensive analyses, this study provides valuable insights into the selection of optimal fine-tuning strategies and elucidates the internal mechanisms underlying multiple cross-domain paradigms. To investigate the intricate trade-off between recognition accuracy and computational cost, we applied nonlinear least squares to construct the Accuracy-Cost trade-off functions. Experimental findings indicated that the optimal transfer learning ratios for these cross-domain paradigms ranged from 6.1% to 9.0%, with most clustering around 9.0%, offering a valuable reference for determining optimal transfer learning ratios within diverse cross-domain scenarios. Additionally, we implemented a real-time online gesture recognition system, validating the feasibility of our approach through preliminary tests in real-world scenarios. In conclusion, this study serves as a preliminary investigation into the application of wrist-worn single-site sensing for fine-grained gesture recognition.

As the core component of precision liquid fertiliser application systems, the anti-clogging mechanisms of liquid fertiliser distributors remain critical theoretical deficiencies. This study employed computational fluid dynamics (CFD) and discrete element method (DEM) to construct the flow fields of solid-liquid two-phase flow within the distributor. The accuracy of the simulation was validated through experimental testing. Based on the simulation process, by analysing the reasons for the formation of the flow fields, the study investigates the motion characteristics of liquid fertiliser in rotating flow fields and at fertiliser outlets under different rotational speeds. The basic movement trend of fibre particles was expounded. The mechanism by which fibre particles are smoothly discharged from the distributor was explored. The results indicate that the velocity distribution within the distributor's internal flow field is inversely correlated with the rotational radius. Across 180-540 rpm, negativepressure vortices can influence pulse-boosting effects within the flow field and the adsorption of particles onto baffles. The outlet pressures formed four synchronised groups (1,5,9/2,6,10/3,7,11/4,8,12), with pulse delays decreasing from 0.10 s (180 rpm) to 0.03 s (540 rpm) as speed increased, simultaneously enhancing synchronicity. The discharge of particles depends on the magnitude of velocity fluctuations, which can be modulated through rotor speed adjustments to prevent the retention of fibre particles and enhance discharge efficiency. These findings provide theoretical support for the design of high-viscosity fluid rotational distribution devices and have engineering guidance value for improving the anti-clogging performance of precision fertilisation equipment.

China relies heavily on imported soybeans due to insufficient domestic production, but these imports are often contaminated with quarantine weed seeds such as A. artemisiifolia and A. trifida. The introduction of these species poses serious ecological risks, highlighting the urgent need for reliable real-time detection methods. In this study, a single-seed uniform distribution and spreading device was designed to minimise occlusion and ensure consistent seed visibility. The device integrates a parabolic seed-socket distribution unit with an embedded system. After seeds were arranged in a single layer on a conveyor belt, a detection camera captured images that were processed by the YOLO_P2 model for seed recognition and counting. Device performance was optimised using the Taguchi experimental design, and evaluated with signal-to-noise ratio, mean, and variance. Experimental analysis revealed that the speeds of the seed-spreading roller and conveyor motor were the most significant factors affecting distribution uniformity. Validation experiments showed that the optimised system achieved detection accuracies of 95.73% for A. trifida and 94.41% for A. artemisiifolia, with an average processing time of 7.6 minutes per sample. These results demonstrate that the proposed device provides a practical, cost-effective solution for quarantine inspection, combining high-throughput capability with real-time performance to support ecological protection efforts.

Increasing market and labour costs arouse an urgent need to develop automatic mechanised tea picking equipment over mountainous tea plantation area. To address the poor operational stability of existing machines, this paper proposes a new design of an omnidirectional dynamic four-point levelling platform and the levelling control algorithms based on four-point levelling and central immobile levelling method. A prototype is built to test the static and dynamic levelling performance on a simulated mountainous terrain. Different from other conventional four-point levelling platforms, the rotary motion is added to this platform and the number of control units is reduced to reduce the difficulty of manoeuvring. These developments result in a damage-free feature when the rotation or inclination motion is beyond the normal manoeuvre range. The test results of the prototype showed that the inclination error of the platform was controlled at 1.1 degrees-1.9 degrees in 2-2.5 s during the static levelling test on a mountainous terrain with a slope of 15 degrees. The dynamic levelling was tested on a mountainous terrain with varied slopes between 10 degrees and 20 degrees and the inclination error range of the platform was 1.7 degrees-2.2 degrees. The field tests confirmed that the levelling system can achieve omnidirectional dynamic levelling in the case of tea picking operations in mountainous areas with slopes below 20 degrees.

In the agricultural field, intelligent picking has great application potential, among which efficient path planning is the key to improving picking efficiency. Given the expansive area and intricate terrain commonly encountered in orchards, challenges such as the tipping over of picking robots at bends and road encroachment, which can damage orchard plants, are prevalent. This study proposes an improved Hybrid A* algorithm based on collaborative path planning for multiple-picking robots. This study focuses on four-wheel steering picking robots and introduces a three-stage path-planning approach based on numerical optimization to address the issue of road encroachment at bends. Furthermore, the efficiency of the algorithm is enhanced by the introduction of guidelines. The improved Hybrid A* algorithm is combined with a time window model to dynamically adjust the priority of picking robots based on the requirements of the picking task and the distance between the robot and the endpoint, thus realizing the collaborative planning of multiple robots and effectively improving the picking efficiency. Finally, the simulation experiment based on MATLAB shows that the algorithm can generate optimal paths for multiple picking robots in orchard environments, adapt swiftly to environmental changes, and exhibit minimal collisions with orchard plants (13.4%). This research not only optimizes the path planning of intelligent picking but also provides strong technical support for the efficient and safe operation of agricultural robots in complex environments.

Precision agriculture was proposed in the 1990s [...]

Quarantine weed seeds in imported soybeans pose significant ecological and economic risks, necessitating highly accurate differentiation between quarantine and non-quarantine weed species. However, conventional manual identification methods are often slow and susceptible to human error, particularly due to the high morphological similarity among different weed seed species. This study proposes an integrated system that combines a novel multi-stage dynamic sorting apparatus with a lightweight deep learning framework, QseedNet, for the automated detection of quarantine weed seeds in imported soybeans. The proposed device physically separates weed seeds from samples, alleviating occlusion and improving imaging quality. QseedNet introduces three key architectural innovations: a Micro-scale Perception Head (MPH) for enhanced detection of ultra-small targets, structural re-parameterization (REP) based on RepVGG for reduced inference complexity, and a Multi-Scale Channel Attention (MSCA) module for improved feature discriminability. Extensive experiments on the Qseed-12 dataset, which includes 12 representative quarantine weed species, demonstrate that QseedNet achieves state-of-the-art performance with a precision of 92.0 %, recall of 90.4 %, and mAP@50 of 95.1 %, while maintaining a compact model size of 3.41 MB. Real-world deployment tests on the NVIDIA Jetson Orin NX platform confirm that QseedNet delivers real-time inference at 70.42 FPS without loss of accuracy. These results highlight the model's suitability for edge-based agricultural inspection workflows.

In this study, a target spraying decision and hysteresis algorithm is designed in conjunction with deep learning, which is deployed on a testbed for validation. The overall scheme of the target spraying control system is first proposed. Then YOLOv5s is lightweighted and improved. Based on this, a target spraying decision and hysteresis algorithm is designed, so that the target spraying system can precisely control the solenoid valve and differentiate spraying according to the distribution of weeds in different areas, and at the same time, successfully solve the operation hysteresis problem between the hardware. Finally, the algorithm was deployed on a testbed and simulated weeds and simulated tillering wheat were selected for bench experiments. Experiments on a dataset of realistic scenarios show that the improved model reduces the GFLOPs (computational complexity) and size by 52.2% and 42.4%, respectively, with mAP and F1 of 91.4% and 85.3%, which is an improvement of 0.2% and 0.8%, respectively, compared to the original model. The results of bench experiments showed that the spraying rate under the speed intervals of 0.3-0.4m/s, 0.4-0.5m/s and 0.5-0.6m/s reached 99.8%, 98.2% and 95.7%, respectively. Therefore, the algorithm can provide excellent spraying accuracy performance for the target spraying system, thus laying a theoretical foundation for the practical application of target spraying.

A lightweight vision-based model, GENet, is proposed to overcome the limitations of conventional missing-seed detection systems, which are highly sensitive to seed characteristics and constrained by slow response and complex configuration. Deployed on a small precision seeder featuring an oblique hook-shaped spoon-type metering device, GENet integrates Ghost Modules, C3Ghost structures, and an ECA attention mechanism. Experiments demonstrate an mAP50–95 of 85.2%, accuracy of 99.9%, and 185 FPS inference speed on the Jetson AGX Xavier platform, while reducing model parameters by over 40%. Validation on the JPS-12 test bench confirms its robustness, providing an efficient solution for intelligent precision seeding.

Traditional passive obstacle avoidance mechanical weeding strategies heavily relied on touch rods, which led to a high crop damage rate and low weeding efficiency during operations. This study proposed an obstacle avoidance information collection scheme that integrates precise detection of obstacle positions and coordinate conversion of weeding tool positions. An active obstacle avoidance control system based on obstacle positions and real-time tool status was designed. This system consisted of the autonomous navigation equipment, obstacle avoidance information collection units, the control system module, hydraulic execution components, and the real-time monitoring sensor. Based on the requirements for active obstacle avoidance, the study established the relationship between the obstacle avoidance information collection units, hydraulic execution components, and the real-time monitoring sensor, and determined a precise active obstacle avoidance control scheme. Field tests were conducted using machine forward speed as the test factor, with inter-row weeding coverage rate and plant damage rate as evaluation indicators. The test results indicated that when the machine forward speed was 460 mm/s, the combined effect of inter-row weeding coverage and operational efficiency was optimal, with an average inter-row weeding coverage rate of 94.62% and a plant damage rate of 1.94%. The active obstacle avoidance weeding scheme proposed in this study provided a technical reference for improving inter-row weeding effectiveness in orchards.

This study addresses the intelligent transformation needs of mine material transportation by proposing an integrated intelligent system incorpora-ting high-precision scene perception technologies.A laser scanning positioning system combined with filtering algorithms achieves dynamic monitoring of raw material pools(error＜5％).A multi-algorithm fusion crane positioning model(positioning accuracy±5 mm)and anti-swing control technology(swing amplitude＜5 °)are developed,supported by 5G-enabled transmission and digital twin models for real-time monitoring.The system integrates UWB positioning(15 cm accuracy),infrared barriers and dynamic digital fencing to establish human-machine collaborative safety mechanisms.Experi-mental results demonstrate 20％improvement in crane operation efficiency,personnel positioning errors＜6 cm,and 35％increase in scheduling re-sponse speed.The research validates the synergistic effectiveness of AI and 5G technologies in complex industrial scenarios,providing a multi-technology integration paradigm for mine intelligence.Future research could integrate edge computing to enhance adaptability in dynamic environment.

To address the issue of increased fuel consumption and reduced efficiency caused by excessive slip of the drive wheels during tractor ploughing operations, this paper considered the time-varying, uncertain, and highly nonlinear characteristics of the tractor-operating unit. A nonlinear dynamic model was constructed and a nonlinear slip control method for the drive wheels was designed using sliding mode variable structure control (SMVSC). The method was validated and tested on both the MATLAB/Simulink platform and a hardware-in-the-loop (HIL) simulation platform based on dSPACE. The HILS results indicated that, compared to the fuzzy PID algorithm, under varying soil specific resistance pulses, the mean absolute deviation of slip rate was reduced by 0.013, and the response time decreased by approximately 1.3 seconds with the SMVSC method. In case of pulse variation in slip rate, the SMVSC method reduced the tracking response time by approximately 0.8 seconds and the average control overshoot by about 0.03. Under both experimental conditions, the SMVSC method demonstrated superior control performance, ensuring more stable tractor operation. These findings provide valuable insights for drive slip control in tractor ploughing operations.

In order to solve the reliability problem of requirement identification at the early stage of the product, a requirement prioritization analysis method based on online review data and entropy weight TOPSIS is proposed for product requirement mining and requirement analysis. Firstly, online review data of the product is collected by web scraper software, then text processing and LDA topic clustering are carried out to identify user requirements; secondly, after considering various factors to formulate suitable design objectives, entropy weight TOPSIS method is used to carry out priority analysis of requirements under multiple objectives; finally, the reliability analysis process of product requirements is completed taking rowing machine as an example. The study demonstrates that requirement prioritization analysis in the initial stages of product development, while considering multi-objective factors, improves subsequent product analysis and design by facilitating a more focused and effective approach.

Robots picking apples cause vibration of other fruit on the same branch and fruit may even fall off. This not only affects subsequent picking, but also causes fruit loss. In this research, the branch-fruit system was taken as the research object, and a mixed-mode cohesive contact model was used to simulate the tangential and directional mechanical behaviour of fruit detachment. The branch with apples was simplified as a cantilever beam model, and a multi-body finite element model of a branch with apples was established. The average error of the simulation results was 6.81%. The simulation results demonstrate that the type I picking sequence (picking fruit beginning farther from the trunk) shortened the settling time of the branch with apples than the type II picking sequence (picking fruit beginning closer to the trunk). Field experiments showed that the maximum accelerations of fruit detachment were respectively 3.826 +/- 1.859g, 2.971 +/- 1.374g, 2.246 +/- 0.757g (g = 9.8 m s-2) under three picking patterns (pulling, pull-twisting, and pull-bending). Linear correlations were found between the vibration displacement and settling time of fruit under different picking patterns and sequences. The response index (RI) is defined to comprehensively evaluate the degree of disturbance to other fruits on the same branch during the fruit-picking process. The test results reveal that the use of the type I sequence and pull-bending pattern is better for reducing the overall disturbance to other apples on the same branch. This study provides a technical and theoretical reference for designing the picking pattern and planning the picking sequence of robotic arms.

To explore the influence of the lateral sloshing and the time -varying mass of the liquid in the tank on the ride comfort of the high -clearance sprayer, a spring -mass -damping equivalent mechanics that can describe the lateral sloshing of the liquid under different filling ratios was constructed based on the equivalent criterion. The Fluent was used to simulate the moment acting on the wall of the tank by the lateral sloshing of the liquid, and then the parameters of the equivalent mechanical model are obtained by fitting and solving. Comparative analysis of Fluent simulation and bench test on lateral sloshing of tank liquid under different filling ratios. The results show that the lateral sloshing trend of the tank liquid level obtained from the Fluent simulation and the bench test was consistent, which proved the accuracy of the Fluent fluid simulation process and the correctness of the required equivalent mechanical model parameters. Incorporating a liquid sloshing equivalent model, a fourdegree -of -freedom vertical dynamic model of the sprayer half -car was established. Subsequently, the performance of the sprayer was systematically analyzed and compared under the excitation of a bump road and a random E -level road. This investigation took into account varying liquid filling ratios of 10%, 50%, and 90%. The focus lay on evaluating the vertical acceleration of the sprayer body, dynamic deflection of the suspension, and dynamic load on the tires in response to these road conditions. This analysis is conducted independently of the liquid sloshing factor. The results show that the lateral sloshing of the liquid medicine significantly reduces the ride smoothness of the machine, and makes the vibration response of the machine produce a certain hysteresis effect. With the reduction of the quality of the liquid medicine in the spray tank, the vibration amplitude of the sprayer body gradually decreases, the hysteresis effect is also gradually weakened. The results presented in this study offer a theoretical foundation for the analysis of ride comfort and the optimization of chassis structure in highclearance sprayers.

Discovering fraud patterns from numerous user activities is crucial for fraud detection. However, three factors make this task quite challenging: Firstly, previous research usually utilize just one of the two forms of user activity, namely sequential behavior and interaction relationship, leaving much information unused. Additionally, nearly all works merely study on a single view of user activities, but fraud patterns often span across multiple views. Moreover, most existing models can only handle regular time intervals, while in reality, user activities occur with irregular time intervals. To effectively discover fraud patterns from user activities, this paper proposes MSTAN (Multi-view Spatio-Temporal Aggregation Network) for fraud detection. It addresses the above problems through three phases: (1) In short-term aggregation, SIFB (Sequential behavior and Interaction relationship Fusion Block) is employed to integrate sequential behavior and interaction relationship. (2) In view aggregation, 2-dimensional multi-view user activity embedding is obtained for simultaneously mining multiple views. (3) In long-term aggregation CTLSTM (Convolutional Time LSTM) is designed to deal with irregular time intervals. Experiments on two real world datasets demonstrate that our model outperforms the comparison methods.