Harvesting robots are often required for the rapid and accurate detection of pear fruit growth quality with real-time picking-point localization under complex orchard environments in smart agriculture. In this study, an improved YOLOv8n model was proposed for the pear fruit detection and localization, particularly for the high accuracy under occlusion and variable illumination, while the high inference speed on embedded devices. Three enhancements were incorporated: (1) The C2f modules in the backbone network were replaced with FasterNet Block using partial convolution. Computational redundancy was significantly reduced to optimize memory access efficiency. (2) A global attention mechanism was introduced after the spatial pyramid pooling fast (SPPF) layer. Critical feature was extracted to focus more effectively on small targets, while suppressing background interference. (3) The original CIoU loss function was replaced with Inner-CIoU using a scale factor of 0.8 after systematic experimentation. The convergence of the improved model was accelerated to enhance the gradient and localization precision for small and overlapping pear fruits. An image dataset was also constructed to verify the improved model. Pear fruit images were captured by an Intel RealSense D455i binocular camera in natural orchards. Multiple varieties and challenging conditions were covered, such as diseases, fruit overlaps, and occlusion. Data augmentation also expanded the dataset to 3,000 images. Experimental results demonstrate that the improved YOLOv8n-Pear model achieved a precision of 96.8%, a recall of 93.4%, and a mean average precision of 96.7%. Compared with the baseline YOLOv8n, these metrics were improved by 4.0, 3.2, and 4.0 percentage points, respectively. Moreover, the floating-point operations were reduced by 30.23% and memory footprint by 48.15%, from 7.1 to 4.2 MB. On the embedded Jetson Orin NX platform, the better performance achieved an average inference speed of 180.3 frames per second with a power consumption of only 19 W, indicating the real-time deployment on power-constrained systems. The binocular camera was calibrated for 3D localization. A coordinate transformation was established to convert 2D pixel coordinates of healthy pear fruits into 3D world ones. Field tests show that the maximum positioning errors in the X, Y, and Z directions were 12, 12, and 10 mm, respectively, with average errors of 6.6, 7.1 and 7.1 mm, respectively, all within acceptable limits for robotic harvesting. Finally, the vision system was integrated with a four-degree-of-freedom harvesting actuator on outdoor Y-trellis pear trees. The better performance was achieved in the harvesting success rate of approximately 90.2% and an average continuous picking time of about 5 s per fruit over ten experimental groups with 100 picking times, fully meeting the practical requirements of robot harvesting. The improved YOLOv8n model effectively balanced the high accuracy and low computational cost. The finding can also provide a robust solution for visual perception in fruit harvesting robots, particularly under resource-constrained embedded platforms.

As the largest fruit producer in the world, China’s comprehensive orchard mechanization rate is below 30% and faces the problem of aging orchard farmers. Thinning is an essential agronomic practice in orchard management. Therefore, to get a marketable product, artificial hand fruit thinning (AHFT) has become a major but costly management practice in modern orchard planting. The authors developed two types of new orchard blossom thinners: a tractor-mounted three-arm blossom thinner (TTBT) and a hand-held electric blossom thinner (HEBT). The arm shape, spindle rotation speed, and rope arrangement density of TTBT can be adjusted flexibly according to the canopy structure of the fruit tree. HEBT is portable and suitable for different canopy types, especially for traditional orchards with a complex-structured canopy. In this paper, a performance evaluation of the two types of blossom thinners on Y-trellis ‘Sucui’ No.1 pear orchard was carried out. In field tests, three treatments were designed and tested, which are TTBT combined with AHFT, HEBT combined with AHFT, and AHFT only. Four indices were used to evaluate the tests: blossom retention rate, fruit setting rate, fruit yield and quality, and work efficiency and cost. The test results showed that the blossom retention rate of TTBT and HEBT at 50% for Y-trellis ‘Sucui’ No.1 pear orchard was perfect; the difference in blossom retention rate and coefficient of variation of every layer of TTBT was very small, and the mean coefficient of variation was 2.97%, which is 1.98% lower than that of HEBT, meaning that the working stability of TTBT was higher than HEBT. The working efficiencies of TTBT and HEBT were much higher than that of AHFT, specifically, 130 and seven times higher, respectively. Although mechanical blossom thinning reduces the fruit setting rate to a certain extent, it has no effect on fruit yield and quality after fruit thinning for final marketable fruit. The profitable areas of TTBT and HEBT were 0.87 hm² and 0.08 hm², respectively.

Due to its ability to broaden the transport channel of droplets within the plant canopy and enhance their penetration capacity, air-assisted spray technology is widely used in orchard pesticide application. To achieve uniform distribution of pesticide droplets in the tree canopy and obtain a higher pesticide utilization rate, it is crucial to clarify the coupling mechanism of the airflow field and droplet field generated by the air-assisted sprayer. This paper introduces a three-dimensional modeling method of the fruit tree canopy based on CFD (Computational Fluid Dynamics), offering a theoretical basis for analyzing the airflow demand calculation during different growth periods of the canopy. It also examines the interaction between canopy modeling and airflow, highlighting advancements in airflow regulation equipment and the effects of airflow speed and volume on spraying. The study shows that the precise regulation of airflow velocity and discharge rate is of importance for improving spraying efficiency. It finally points out that future research should focus on developing intelligent regulation equipment for efficient airflow-droplet control, using biomass sensing, which involves measuring the growth characteristics of the tree canopy, to meet the needs of orchards with diverse growth stages and canopy structures. This article could provide guidance for the future study of precision air-assisted spraying technology in orchards.

To solve the problem of effective utilization of orchard green fertilizer, a small one-rotor orchard horizontal rotary rake (OHRR) was developed, which is used for grass collection in the mowing agronomic section of orchard management. In the working process, the disc drives the arms rotating around the vertical shaft, the guide cam that is fixed to the vertical shaft controls tine movement, and the tines turn to different inclination angles in different positions. The kinematic validation model of OHRR was built based on the theory that there is no gap or small overlap between the two adjacent working areas of the tines. This model determines the relationship between the advancing speed, disc rotational speed, rotation radius, arm number, and tine working width. The leakage and repeating rate of OHRR virtual prototype were calculated by tines movement trajectories analysis in multi-body dynamics simulation. Box-Behnken three-factor and three-level test plans for advancing speed, disc rotational speed, and tine working width were designed to obtain the optimal operation parameters of the OHRR: advancing speed was 11.16 km/h, disc rotational speed was 6.98 rad/s, and tines working width was 0.3 m. Taking labor working as the control group, OHRR field tests were evaluated by four indices: strip density, leakage rate, working efficiency, and profitable area. Field tests results showed that the leakage rate of OHRR was 4.56%, which meets the requirements of national standard JB/T10905. The strip density, width, and height of OHRR were 29.44 kg/m³, 0.5 m, and 0.25 m, respectively. These data can provide support for the subsequent loading and transportation operation. The profitable area of OHRR was 6.7 hm², which is suitable for large-scale mechanized orchard management.

With the rapid advancement of modern agriculture, mechanized and intelligent pollination has emerged as a crucial focus for enhancing agricultural efficiency and minimizing labor expenses. Traditional pollination methods, limited by environmental factors and high labor costs, fail to adequately address the production demands of large-scale orchards and vegetable gardens. Consequently, researchers have integrated mechanized equipment, drone technology, robotics, and deep learning algorithms to achieve accurate identification and precise pollination on inflorescences. The research on mechanized and intelligent pollination has not only injected new momentum into the field of fruit and vegetable pollination but also provided key technological support for addressing global agricultural labor shortages and increasing crop yields. This review summarizes recent advances in mechanized and intelligent pollination, focusing on deep learning’s role in object recognition, improvements in pollination equipment, and the effectiveness of intelligent pollination across various fruits or vegetables. Studies indicate that mechanized and intelligent pollination significantly enhances working efficiency and fruit yields, though it continues to face challenges such as technical complexity and high implementation costs. Looking ahead, as robotics and artificial intelligence algorithms continue to advance, mechanized and intelligent pollination is poised for broader adoption in agricultural management practices. This review systematically summarizes the research progress in mechanized and intelligent pollination technologies for fruit and vegetable crops, providing significant theoretical support and reference value for future studies in crop pollination techniques.