The sampling-based RRT* algorithm has been extensively employed for path planning. The improved variant, F-RRT*, significantly reduces the initial cost by generating new nodes; however, it also leads to increased computational time and poses challenges in narrow passage environments. To address these issues, this paper proposes an improved F-RRT* algorithm based on the Generalized Voronoi Diagram (GVD), called GVDF-RRT*. First, a sparse sampling strategy based on GVD is proposed, which improves the sampling efficiency and reduces the initial sampling nodes and initial time by guiding the random tree to explore the space quickly through GVD nodes. At the same time, the success rate of planning in narrow passage environments is improved by equalizing the sampling probability of each channel in the map. Secondly, a node direct connection mechanism is established, which reduces the redundant computation and further reduces the initial time by hierarchical processing of creation nodes, while combined with the forward optimization mechanism to optimize the paths in iteration, effectively reducing the path cost. Finally, simulations in three test environments show that, compared to RRT*, Q-RRT*, and F-RRT*, the GVDFRRT* algorithm reduces the initial number of sampled nodes by 47.42%-76.57%, the initial time by 37.07%-86.00%, and the initial path cost by 0.47%-0.91%, while significantly improving the success rate in narrow passage environments. This further demonstrates that the proposed GVDF-RRT* significantly outperforms the baseline method in terms of initial performance and adaptability to narrow passage environments.

Inventory counting of logistics boxes in complex scenarios has always been a core task in intelligent logistics systems. To solve the problems of a high leakage rate and low computational efficiency caused by stacking, occlusion, and rotation in box detection against complex backgrounds in logistics environments, this paper proposes a lightweight, rotated object detection model: REW-YOLO (RepViT-Block YOLO with Efficient Local Attention and Wise-IoU). By integrating structural reparameterization techniques, the C2f-RVB module was designed to reduce computational redundancy in traditional convolutions. Additionally, the ELA-HSFPN multi-scale feature fusion network was constructed to enhance edge feature extraction for occluded boxes and improve detection accuracy in densely packed scenarios. A rotation angle regression branch and a dynamic Wise-IoU loss function were introduced to further refine localization and balance sample quality. Experimental results on the self-constructed BOX-data dataset demonstrate that the REW-YOLO achieves 90.2% mAP50 and 130.8 FPS, with a parameter count of only 2.18 M, surpassing YOLOv8n by 2.9% in accuracy while reducing computational cost by 28%. These improvements provide an efficient solution for automated box detection in logistics applications.

To address the issues of excessive sheet metal thinning and geometric deviation in single point incremental forming (SPIF), this paper proposed a bi-objective process parameter optimization framework for Al1060 sheet based on a multilayer perceptron (MLP) surrogate model and an improved multi-objective grey wolf optimization (IMOGWO) algorithm. Finite element simulations based on ABAQUS were conducted to generate a dataset considering variations in tool radius, initial sheet thickness, tool path strategy, step depth and forming angle. The trained MLP was used as the objective function in the optimization process to enable the rapid prediction of forming quality. The IMOGWO algorithm, enhanced by the Spm chaotic mapping initialization, an improved convergence coefficient updating mechanism and associative learning mechanism, was then employed to efficiently search for Pareto optimal solutions. For a truncated conical component case, optimal parameter sets were selected from the Pareto front via the entropy-weighted TOPSIS method for order preference by similarity to an ideal solution. Experimental verification showed close agreement with the simulated results, with relative errors of only 0.58% for the thinning rate and 3.10% for the geometric deviation. This validation demonstrates the feasibility and potential of the proposed method and its practical potential for improving the quality of SPIF forming.