Efficient obstacle-avoidance path planning is critical for orchards with numerous irregular obstacles. This paper presents a continuous bidirectional Quick-RRT* (CBQ-RRT*) algorithm based on the bidirectional RRT (Bi-RRT) and Quick-RRT* algorithms and proposes an expansion cost function that evaluates path smoothness and length to overcome the limitations of the Quick-RRT* algorithm for non-holonomic mobile robot applications. To improve the zigzag between dual trees caused by the dual-tree expansion of the Bi-RRT algorithm, CBQ-RRT* proposes the CreateConnectNode optimization method, which effectively solves the path smoothness problem at the junction of dual trees. Simulations conducted on the ROS platform showed that the CBQ-RRT* outperformed the unidirectional Quick-RRT* in terms of efficiency for various orchard layouts and terrain conditions. Compared to Bi-RRT*, CBQ-RRT* reduced the average path length and maximum heading angle by 8.5% and 21.7%, respectively. In addition, field tests confirmed the superior performance of the CBQ-RRT*, as evidenced by an average maximum path lateral error of 0.334 m, a significant improvement over Bi-RRT* and Quick-RRT*. These improvements demonstrate the effectiveness of the CBQ-RRT* in complex orchard environments.

Autonomous mobile robots play a vital role in the mechanized production of orchards, where human-following is a crucial collaborative function. In unstructured orchard environments, obstacles often obscure the path, and personnel may overlap, leading to significant disruptions to human-following. This paper introduces the KCF-YOLO fusion visual tracking method to ensure stable tracking in interference environments. The YOLO algorithm provides the main framework, and the KCF algorithm intervenes in assistant tracking. A three-dimensional binocular-vision reconstruction method was used to acquire personnel positions, achieving stabilized visual tracking in disturbed environments. The robot was guided by fitting the personnel’s trajectory using an unscented Kalman filter algorithm. The experimental results show that, with 30 trials in multi-person scenarios, the average tracking success rate is 96.66%, with an average frame rate of 8 FPS. Additionally, the mobile robot is capable of maintaining a stable following speed with the target individuals. Across three human-following experiments, the horizontal offset Error Y does not exceed 1.03 m. The proposed KCF-YOLO tracking method significantly bolsters the stability and robustness of the mobile robot for human-following in intricate orchard scenarios, offering an effective solution for tracking tasks.

Precise localization of the fruit stem picking point is crucial for robots to achieve efficient harvesting operations. However, in unstructured orchard environments, citrus fruit stems are easily obscured by branches and leaves and affected by factors such as overlapping fruits. This leads to poor picking localization accuracy for robots, impacting their autonomous picking efficiency. Therefore, this paper proposes a method for estimating the posture of citrus fruit stems and performing picking operations under environmental occlusion, based on the YOLO-OBB algorithm. First, the YOLOv5s algorithm detects the ROI of citrus, combined with depth information to obtain their 3D point clouds. Second, the OBB algorithm constructs oriented point cloud bounding boxes to determine stem orientation and picking point locations. Finally, through hand–eye pose transformation of the robotic arm, the end-effector is controlled to achieve precise picking operations. Experimental results indicate that the average picking success rate of the YOLO-OBB algorithm reaches 82%, representing a 50% improvement over approaches without fruit stem estimation. This conclusively shows that the proposed algorithm provides precise fruit stem pose estimation, effectively enhancing robotic picking success rates under constrained fruit stem detection conditions. It offers crucial technical support for autonomous robotic harvesting operations.

The identification of defects in micro coaxial cable harnesses is of paramount importance in ensuring the quality of the production process. However, the prevailing methodology for detecting such defects in micro coaxial cable harnesses is predominantly manual, which has resulted in poor detection accuracy and precision. Furthermore, this approach had a deleterious effect on both the efficiency and quality of the production process. In order to achieve a rapid and precise defect detection technology, this paper provided a comprehensive analysis of the production line status and the various defect types associated with micro coaxial cable harnesses. Subsequently, the application of traditional image processing methods, machine learning methods, and deep learning methods to defect detection technology for high-precision components was discussed, and summarized the algorithmic performance, advantages and limitations of the various methods were summarized. The findings of this paper demonstrated that traditional image processing methods and machine learning methods exhibited inadequate robustness, were susceptible to illumination-related variations, and had a restricted scope of applicability. In contrast, deep learning methods demonstrated enhanced detection precision and efficiency, nevertheless, remaining constrained by certain limitations. This paper presents a discussion and proposes the corresponding suggestions from three perspectives: the impact of illumination environment, defect dataset and defect detectability. The objective was to enhance the precision and robustness of visual defect detection, thus establishing a foundation for the subsequent design of visual inspection system for defects in micro coaxial cable harnesses. Furthermore, it furnished pertinent relevant references for the application of high-precision component defect visual inspection.

As an important component of photovoltaic power generation, PV panels play a crucial role in the photovoltaic power generation industry. In order to overcome the current problem of low speed and accuracy in detecting hot spot faults of PV panels in photovoltaic power plants, this paper proposes a lightweight YOLO V5 model to realize the detection of hot spot defects of PV panels in infrared images. First, data enhancement is performed on the existing dataset so as to increase the number of samples i n the database and enhance the generalization ability of the model. Second, a lightweight network structure Shufflenet v2 is introduced to replace the YOLO V5 backbone network to enhance the speed of detection. Finally, the attention mechanism is combined with the neck network to enhance the extraction of hot spot features. It can effectively balance the network's speed and accuracy. The experimental results show that the improved network achieves recognition accuracy of 93.5% for hot spots, which is 4.4% higher compared to the original network.