The rapid scaling of semiconductor technologies has led to increasingly complex integrated circuit layouts, which pose severe challenges to maintaining manufacturing yield and highlight the urgent need for accurate lithographic hotspot detection. Existing detection methods are often constrained by limited layout modelling capacity, insufficient separability of augmented samples, and suboptimal classification performance. To overcome these challenges, a hotspot detection algorithm was developed that combined a multi-scale graph family modelling strategy with a hierarchical graph neural network. The proposed graph family model constructed both local subgraphs and a global graph to represent geometric details and contextual semantics simultaneously, thereby improving structural representation and preserving data separability under various augmentation operations. On this basis, a hierarchical GNN was designed to extract multi-level features and strengthen the model’s focus on critical core regions. Experimental evaluations on the ICCAD'19 TNSB benchmark demonstrate that the proposed method substantially improves data distribution balance, achieving a 206 percentage point increase in MNND and a 16 percentage point gain in Shannon entropy. Moreover, the detection model attains a recall of 99.91% with only a 1.12% false alarm rate, outperforming state-of-the-art alternatives. These results confirm that the proposed graph-based approach offers both strong structural expressiveness and robust detection capability, providing a promising solution for addressing the challenges of lithographic hotspot detection in advanced technology nodes.

Aiming to address low reconstruction accuracy caused by improper selection of matching source in the stereo matching process of left and right images in binocular structured light 3D reconstruction, a matching method based on absolute phase is proposed to reduce the mismatch rate and improve the reconstruction accuracy. Firstly, based on the absolute phase solved by phase shift method and multi-frequency heterodyne method, the polar line constraint is used to create a search matching area by global search and local search for the pixels of the left and right images. Then, cubic spline interpolation is performed on the search area to achieve sub-pixel stereo matching. Finally, the triangulation principle is used to complete the conversion of the two-dimensional coordinates of the matching points to three-dimensional coordinates, and the three-dimensional reconstruction of the target surface is realized. The experimental results show that the method proposed reduces the matching time by 87.77% compared to SAD(Sum of Absolute Differences) and SSD(Sum of Squared Differences) matching algorithms, and improves the reconstruction accuracy by 77.75% and 90%, respectively, demonstrating its strong generality and practicality. Particularly, the point cloud reconstructed in 3D has no water ripple phenomenon, and the surface is smoother and closer to the actual object.

Path planning is the key to realize the autonomous navigation of mobile robots. Aiming at the problems such as low search efficiency of traditional ant colony algorithm, proneness to falling into local optimal and insufficient dynamic obstacle avoidance ability, a path optimization method combining improved ant colony and dynamic window approach (DWA) is proposed to realize global path optimization and improve local dynamic obstacle avoidance ability of mobile robots. In the global path planning, the heuristic function is established by introducing the artificial potential field factor to enhance the orientation of ants to the target point in the process of path searching, so as to accelerate the search speed of the algorithm. Secondly, in the pheromone update strategy, the extra increment of pheromone concentration difference between the optimal path and the worst path in the previous generation is added to adaptively update the pheromone concentration and enhance the global optimization ability of the algorithm. Then, the triangular pruning method is used to delete the redundant transition nodes of the global path and shorten the path length. Finally, cubic B-spline curve is introduced to optimize the path inflection point and improve the path smoothness. In the local path, an obstacle avoidance cost subfunction considering the speed factor is added to the evaluation function of DWA to increase the local dynamic obstacle avoidance ability of the algorithm, so that the robot can detect and avoid obstacles in real time while moving. The simulation results show that the proposed ant colony algorithm with DWA can improve the path length, iteration times, number of turning points and path smoothness compared with the traditional algorithm, and can effectively improve the dynamic obstacle avoidance ability.

Road traffic cones are safety isolation signs used for highway and urban road maintenance. Automated equipment can greatly improve operational efficiency and reduce safety hazards for workers caused by traditional manual retrieval methods. Accurate target detection of traffic cones is key to ensuring the normal operation of automated retrieval equipment. To improve recognition efficiency while ensuring model detection accuracy and reducing model computation, a lightweight road traffic cone detection algorithm called TraCone-YOLO (TraCone-You Only Look Once) is proposed in this paper. Firstly, an efficient detection head based on shared depthwise separable convolution is adopted in this paper to make the model more efficient in inference. Secondly, to improve the model's ability to extract multi-scale feature information, the RepC2f module is used in the feature extraction process. Finally, by introducing the ELA (Efficient Local Attention) attention module, an ELA-HSFPN (ELA-Hierarchical Scale based Feature Pyramid Network) network structure is proposed for multi-scale feature fusion, further enhancing the model's feature expression ability.In response to the lack of samples with different weather and light intensities in the original traffic cone dataset called TraCon, data augmentation method is applied to simulate different weather conditions in actual scenarios. The training samples are effectively expanded, thereby improving the model's generalization ability and robustness. The experimental results show that the improved model reduces parameters and computation by 56.2% and 55.6% respectively, with an average precision of 92.6%, which is only 0.4 percentage points lower than the original model. The algorithm proposed in this research can make model lightweight while maintaining good detection performance.