To enhance the efficiency of mechanical automatic picking of cherry tomatoes in a precision agriculture environment, this study proposes an improved target detection algorithm based on YOLOv5n. The improvement steps are as follows: First, the K-means++ clustering algorithm is utilized to update the scale and aspect ratio of the anchor box, adapting it to the shape characteristics of cherry tomatoes. Secondly, the coordinate attention (CA) mechanism is introduced to expand the receptive field range and reduce interference from branches, dead leaves, and other backgrounds in the recognition of cherry tomato maturity. Next, the traditional loss function is replaced by the bounding box regression loss with dynamic focusing mechanism (WIoU) loss function. The outlier degree and dynamic nonmonotonic focusing mechanism are introduced to address the boundary box regression balance problem between high-quality and low-quality data. This research employs a self-built cherry tomato dataset to train the target detection algorithms before and after the improvements. Comparative experiments are conducted with YOLO series algorithms. The experimental results indicate that the improved model has achieved a 1.4% increase in both precision and recall compared to the previous model. It achieves an average accuracy mAP of 95.2%, an average detection time of 5.3 ms, and a weight file size of only 4.4 MB. These results demonstrate that the model fulfills the requirements for real-time detection and lightweight applications. It is highly suitable for deployment in embedded systems and mobile devices. The improved model presented in this paper enables real-time target recognition and maturity detection for cherry tomatoes. It provides rapid and accurate target recognition guidance for achieving mechanical automatic picking of cherry tomatoes.

The automated harvesting of clustered fruits relies on fast and accurate visual perception. However, the obscured stem diameters via leaf occlusion lack any discernible texture patterns. Nevertheless, our human visual system can often judge the position of harvesting points. Inspired by this, the aim of this paper is to address this issue by leveraging the morphology and the distribution of fruit contour gradient directions. Firstly, this paper proposes the calculation of fruit normal vectors using edge computation and gradient direction distribution. The research results demonstrate a significant mathematical relationship between the contour edge gradient and its inclination angle, but the experiments show that the standard error projected onto the Y-axis is smaller, which is evidently more conducive to distinguishing the gradient distribution. Secondly, for the front view of occluded lychee clusters, a fully convolutional, feature prototype-based one-stage instance segmentation network is proposed, named the lychee picking point prediction network (LP³Net). This network can achieve high accuracy and real-time instance segmentation, as well as for occluded and overlapping fruits. Finally, the experimental results show that the LP³Net based on this study, along with lychee phenotypic features, achieves an average location accuracy reaching 82%, significantly improving the precision of harvesting point localization for lychee clusters.

The primary focus of research in agricultural unmanned aerial vehicle (UAV) pesticide application technology is the investigation of droplet drift and deposition. The influence of the rotor airflow on droplets is particularly significant, making numerical simulations a crucial tool for airflow field analysis. Among existing numerical simulation methods, the Finite Volume Method (FVM) and the Lattice Boltzmann Method (LBM) are commonly used, but there is limited research that compares the two approaches. Therefore, this paper conducts numerical simulations of the rotor airflow of an agricultural UAV using Fluent, representing the FVM, and XFlow, representing the LBM. This research aims to reveal the distribution patterns of airflow field numerical simulations under different theoretical methods, validate them through practical experiments, and select the optimal method for simulating rotor airflow. The ultimate goal is to establish an effective airflow field model to enhance the precision of pesticide application by an agricultural UAV. The results indicate that the lift error calculated by XFlow in this paper is 2.57% smaller than that by Fluent. The wind field of Fluent entered the “stable state” earlier than XFlow, and the speed value of Fluent was smaller than that of XFlow. The difference between the two speed values became larger and larger as the distance from the rotor was longer. Compared with XFlow, Fluent changes more obviously in the core region, and the center region gradually disappears with the distance from the rotor. However, in the velocity field calculated by XFlow, there are still more turbulent flows outside the core region, indicating that the transient calculation method based on the LBM can better show the details of fluid flow than the steady-state calculation method based on the FVM. Through comparison with the actual test data, it is found that the relative error of the velocity value of XFlow at 0.2 m and 0.4 m is small, while that of Fluent at 0 and 0.2 m is small. It shows that XFlow simulation has higher accuracy for external turbulent flow, while Fluent simulation has higher accuracy for steady laminar flow. The research results provide data comparison and a basis for further analysis of the wind field model of the rotor wing of the plant protection UAV, and they lay a foundation for further research on precision application technology.