Accurate detection of cherry fruits is crucial for improving fruit quality, optimizing harvesting timing, and automating orchard management. However, due to factors such as small fruit size, diverse colors, complex backgrounds, and overlapping occlusions, detection still faced many challenges. To this end, this paper proposed a lightweight and improved model, ACS-YOLOv8n, based on the YOLOv8n architecture to enhance detection accuracy and efficiency. The model introduced an adaptive downsampling module (ADown) in the backbone network, which integrates the advantages of average pooling and max pooling, and combines multi-path convolution operations to effectively retain both global information and local salient features, thereby enhancing the detection capability of small targets. Meanwhile, the Context Guided Block (CGB) was used to replace the traditional C2f module, improving the model's ability to perceive the relationship between the target and the environment and enhancing the effect of multi-scale feature extraction through the joint extraction of local, contextual, and global features. In the neck part, the Slim-Neck structure based on GSConv and VoV-GSCSP was introduced to optimize the feature fusion method, significantly reduce computational redundancy, and balance speed and accuracy. The experimental results showed that ACS-YOLOv8n achieved 92.94% precision, 90.66% recall, and 96.57%mAP50 in complex orchard environments. Compared with YOLOv8n, the precision increased by 2.62 precentage point, the recall by 1.57 precentage point, and the average precision by 2.08 precentage point, while the number of model parameters and the computational volume were reduced by 31.90% and 33.33%, respectively, and the model size was 71.43% of the original model. The proposed model demonstrated good detection performance and robustness in complex environments, providing effective technical support for automatic cherry harvesting and intelligent orchard management.

As the world's largest vegetable producer and consumer, the area of vegetable cultivation in China is constantly increasing. However, it faces challenges such as relatively low levels of mechanization and intelligentization, as well as the annual increase in labor costs. With the development of robot technology, agricultural robots have been increasingly widely applied in the vegetable production field and have become a key technical means to replace manual labor and improve production efficiency and quality. To systematically understand the research status of robotic technology in vegetable production and clarify its future development direction, this paper first elaborates on the definition of agricultural robots. Secondly, focusing on the three core operation links in vegetable production: sowing and transplanting, field management, and harvesting and sorting, it sorts out advanced research results at home and abroad, and analyzes the relevant research progress from three dimensions: perception, decision-making, and control. This includes the SLAM-based autonomous navigation technology and path planning principles based on multi-information fusion; the technical routes and key technology research levels of machine vision application in various links of vegetable production; and the progress achieved in the control technology of operation execution mechanisms and end-effectors in response to specific agronomic requirements and operational environment characteristics. Finally, the paper analyzes the main technical challenges faced by current agricultural robots in vegetable production, namely the significant bottlenecks in applicability and operational efficiency, and looks forward to future development directions. Future research should focus on the development of multi-functional operation robots, which requires breaking through multi-robot collaborative control technology and deepening specialized and modular design; at the same time, policy support is also indispensable. With the continuous advancement of technology and the joint efforts of various parties, robotic technology is expected to become a key support for optimizing the agricultural labor structure, improving vegetable production efficiency, and realizing smart agriculture.

This study takes the patent information of China's electric agricultural machinery technology as the research object, and uses patent measurement analysis method and visual analysis tools to carry out statistical analysis and comparative research on the relevant patent information from the perspectives of annual patent applications, patent structure, technological key, main applicants, patent layout, technology lifecycle and so on. The research results show that China's electric agricultural machinery technology innovation is developing rapidly, and the patent categories are mainly utility model and invention patents, and the proportion of invention patents is lower than utility model patents. Universities are the most important innovation subjects in this field, and technology research and development focuses on IPC (international patent classification) subcategories such as A01C (seeding, fertilization), B60K (power, transmission), and A01D (harvesting, mowing). Jiangsu Province and Shandong Province rank the top 2 in the number of domestic patent applications, and the United States and Japan are the two countries with the most layout in China. At present, the growth of electric agricultural machinery technology is weak, but the new technology has strong features, there is no sign of maturity, and it does not have signs of technological aging and still has greater development potential. The results of the study can provide information reference for clarifying the status quo, characteristics and development trend of China’s electric agricultural machinery technology innovation.

Fish behavior, as a vital sub-discipline of animal behavior, aims to uncover the activity patterns and behavioral mechanisms of fish under both natural and artificial conditions. With the intensification and intelligent transformation of aquaculture, in-depth research on fish behavioral characteristics and their dynamic changes is of great significance for improving farming efficiency, optimizing management strategies, and ensuring fish welfare. This paper systematically reviews the characteristics, influencing factors, and recent research progress of feeding, swimming, stress, and social behaviors in fish. It highlights the application and practice of technologies such as video monitoring, image recognition, behavioral quantification, and deep learning in intelligent behavior recognition and data perception. Current studies have shown that various behavior types are closely related to fish health, growth, and environmental adaptability. Their dynamic changes can serve as critical references for precise feeding, environmental regulation, and disease early warning in aquaculture management. Fish behavior research is evolving from traditional descriptive observation toward intelligent and data-driven approaches. In the future, it is expected to play a key role in deep-sea aquaculture, ecological farming, and the intelligent upgrading of aquaculture equipment. Building behavior-driven intelligent management systems will contribute to achieving more efficient and sustainable aquaculture models.

The conveying of grains is a critical process in the operation of combine harvesters, and the performance of this conveying system directly affects the overall operational efficiency of the harvester. However, traditional screw conveyors often suffer from low conveying efficiency during grain transport, which has become a bottleneck limiting the improvement of harvester performance. To address this issue, this study focuses on optimizing the structural parameters of key components in screw conveyors with the aim of enhancing their conveying performance. A novel discontinuous folded-edge structure was innovatively proposed. Based on the discrete element method (DEM) and orthogonal experiments design, folded-edge width B, axial folded-edge length L_x, and axial spacing between folded-edges S_x were systematically investigated with mass flow rate q as the performance evaluation indicator. The optimal combination of structural parameters was determined through range analysis. To verify the reliability of this optimization scheme, bench tests and high-speed photography were conducted to compare the conveying efficiency and material motion characteristics between the optimized design and the traditional screw conveyor. The research results indicate that, orthogonal simulation experiments identified the optimal parameter combination for the discontinuous folded-edge structure: B=5 mm, L=20 mm, S=0 mm. Bench tests confirmed the reliability of the optimized structure, demonstrating a 5.41% to 7.47% improvement in conveyance efficiency compared to the prototype. High-speed photography analysis further revealed that the discontinuous folded-edge structure effectively alters the flow state of particles, increasing the ratio of axial velocity to resultant velocity. This results in a higher axial conveying speed, enhances the axial transport of grain particles, and reduces energy loss. This study designed a screw conveyor with a discontinuous folded-edge structure, which improved the conveying performance of grain particles. The findings provide a solid theoretical basis and practical guidance for the structural optimization of screw conveyors.

With the development of agricultural science and technology, factory seedling nursery technology has gradually become an important part of modern agricultural production. The introduction of intelligent seeding detection technology significantly improves the efficiency and accuracy of the seedling nursery process and promotes the intelligent and automated process of agricultural production. Intelligent seeding detection is the key to achieving efficient production of factory seedling nursery. Seeding detection technology has gradually changed from photoelectric detection and capacitive detection to visual detection combined with hyperspectral technology and other methods. The applicable objects of photoelectric detection, capacitance detection, visual detection, and hyperspectral detection technology methods are outlined, and the advantages and disadvantages of existing photoelectric, capacitance, and visual detection technologies are summarised. In photoelectric detection researchers generally study three aspects of detection indicators, heavy leakage detection, and improvement of detection accuracy in high dust environments, which are generally applicable to large grain crops such as soybeans, corn, etc. For capacitive detection, the main analysis of capacitance structure on seeding detection, in addition to the optimization aspects of the polar plate structure, algorithm optimization through signal processing to reduce environmental and signal interference, improve detection accuracy and work stability, reducing the error rate. For the visual detection of domestic and international research is mainly aimed at the leakage of seeding, seed flow, seed falling speed seed, detection algorithm model and deep learning algorithm innovation and other directions of research, for the traditional detection technology accuracy is not high, capacitance detection distance is far away from the problem of visual detection error, put forward the development trend of intelligent seeding detection technology for seedlings and nursery seedlings from detecting leakage of seeding to the other seed biochemical properties, and predicting the yield of seedlings and nursery seedlings, and so on. Provide important decision-making basis and reference significance for building the development of intelligent seedling sowing performance detection related technology based on hyperspectral technology.

Camellia Sinensis var. assamica in Hainan serves as a representative and critical local germplasm resource in Hainan, playing a pivotal role in the development of the Hainan tea industry. To further explore its biological resources and economic value, 50 samples of Camellia Sinensis var. assamica germplasm were analyzed based on five quality traits: water extract, caffeine, theanine, tea polyphenols, and the phenol-ammonia ratio. The results indicate that the germplasm resources of Camellia Sinensis var. assamica possess rich genetic diversity. The average coefficient of variation for the five quality traits was 16.87%, with an average diversity index of 3.984. Caffeine exhibited the highest coefficient of variation at 47.44%, while the phenol-ammonia ratio had the lowest at 0.22%. Correlation analysis revealed that the water extract was positively correlated with theanine, tea polyphenols, and the phenol-ammonia ratio, whereas theanine showed negative correlations with tea polyphenols and the phenol-ammonia ratio. Principal component analysis identified two principal components, accounting for a cumulative contribution rate of 79.12%, which reflect key indicators of basic quality and flavor specificity. Suitability analysis demonstrated that the phenol-ammonia ratio of Camellia Sinensis var. assamica is less than 8, making it suitable for green tea production. Notably, the Wanning region is characterized by high theanine content and a low phenol-ammonia ratio, ideal for producing high-quality fresh green tea. Wuzhishan tea exhibits high polyphenol content, showing great potential for black tea and dark tea production. Lingshui Mountain tea has a high caffeine content, suitable for oolong tea development. Cluster analysis grouped the germplasm into four clusters, with the second and fourth clusters (medium and lower-middle levels) comprising 76%. The first cluster (high-content group) and the third cluster (low-content group) can serve as complementary materials for directional breeding. To sum up, Camellia Sinensis var. assamica exhibits significant potential in tea variety breeding, processing technology optimization, and the development of distinctive tea products. In future research, it is essential to integrate genetic marker analysis to further elucidate the genetic basis of its quality traits. Additionally, the quality of tea production should be rigorously evaluated through sensory assessment to ensure precise resource utilization and thereby promote the high-quality development of the Hainan tea industry.

Agricultural robots constitute a crucial trajectory in the evolution of modern agriculture, and has emerged as a focal point of strategic global competition in agricultural equipment technology. As a specialized domain within robotics, agricultural robots are fundamentally composed of an information perception system, decision-making control system, operation execution mechanism, and autonomous mobile platform, having gone through multiple development stages from semi-automation to intelligence. This paper aims to provide a precise understanding of the contemporary research landscape and future trajectories of agricultural robots by delineating their conceptual framework, reviewing relevant national and international development policies, and outlining the underpinning technical background. Agricultural management robots and picking robots are two core directions in the field of agricultural robotics. Management robots achieve precise control of production management operations through real-time monitoring of crop growth parameters and environmental conditions. Picking robots, on the other hand, automatically perform crop picking tasks through autonomous navigation, intelligent recognition and precise control technology. This paper focuses on cutting-edge progress, technical bottlenecks, and key breakthrough directions concerning various management robots, including those for information acquisition, pollination, weed control, plant protection, leaf pruning and intelligent irrigation systems, as well as picking robots dedicated to fruits, vegetables, flowers, and tea. Considering the distinct requirements and characteristics of China's agriculture, this analysis critically analyzed the technological bottlenecks and key breakthrough directions of agricultural robots, including precise information perception and recognition, real-time data processing and intelligent decision-making, adaptive chassis path planning and autonomous navigation, and operational actuator design and precise operational control. It is envisioned that agricultural robots will be widely applied in the fields of crop management, precision irrigation, pest control and picking, thereby accelerating the the evolution of agriculture towards a more intelligent, precise, and autonomous paradigm. Relying on advanced technology to address challenges in agricultural production, it will promote the optimization of agricultural workforce and the improvement of production efficiency, and server as a foundational pillar for the realization of smart agriculture.

The conventional honey production is dominated by fragmented, small-scale individual farming models. The traditional approach of honey-harvesting involving manual beehive frames extraction, beeswax layer excision and centrifugal honey separation, expose beekeepers to potential bee stings and frequently compromise honeycomb integrity. To address these limitations, we designed an automated honey-harvesting robot capable of autonomous frame extraction and beeswax removal. The robot mainly consists of a mobile mechanism equipped with image recognition for beehive localization, a magnetic adsorption-based beehive frame handling device (60.8 N maximum suction) coupled with a cross-slide mechanism for precise frame manipulation, and a thermal beeswax layer-melting apparatus, with optimal melting parameters (15 m/s airflow at 90 ℃ for 30 seconds) determined through rigorous thermal flow simulations utilizing FLUENT/Mechanical software. Field experiments demonstrated beehive frames handling success rate exceeding 85%, beeswax layer removal efficacy over 80% and damage of honeycombs below 30%. The experiment results validate the robot's operational reliability and its capacity to automate critical harvesting procedures. This study significantly reduces the labor intensity for beekeepers, effectively eliminates the risk of direct human-bee contact and improves the removal of beeswax layer, thereby catalyzing the modernization of the beekeeping industry.

Orchard operational scenarios present significant challenges for visual perception, including high vegetation heterogeneity, dynamic lighting variations, and diverse target morphologies. Traditional single-task visual perception models suffer from low feature reusability and high computational redundancy, thereby inadequately addressing real-time environmental perception demands for agricultural robots. This study proposes AgriYOLOP, a lightweight multi-task collaborative perception framework specifically designed for orchard environments. Through a systematic reconstruction of the YOLOP architecture, AgriYOLOP incorporates an efficient backbone network, enhanced anchor-free detection techniques, feature pyramid networks (FPN), path aggregation networks (PAN), and task-adaptive loss function weighting strategies. This framework faciliates parallel collaborative processing of three critical percetption tasks: trunk detection, obstacle recognition, and traversable region segmentation. The proposed framework was validated on a self-constructed orchard dataset comprising 4 765 images (1 280 pixels×720 pixels), captured across diverse seasons, lighting conditions, and vegetation growth stages. Experimental results demonstrate that AgriYOLOP achieves 92.7% precision, 94.6% recall, and 96.7% mAP50 in object detection tasks, along with 98.3% recall, 99.1 F1 score, and 98.1% mIoU in traversable region segmentation. Deployed on an NVIDIA RTX 4060 platform, the model attains 69 f/s real-time inference speed with only 14 M parameters. Comparative experiments reveal that the multi-task collaborative architecture significantly enhances feature-sharing efficiency, reducing inference latency by 32.6% compared to single-task models while improving robustness to illumination and seasonal variations. This approach effectively mitigates the conventional trade-off between target detection accuracy and semantic segmentation efficiency encountered in real-time agricultural robotic applications. The study provides a high-precision, low-latency real-time perception solution for autonomous orchard robot navigation.