Laser vision sensors for weld seam extraction face critical challenges due to arc light and spatter interference in welding environments. This paper presents a real-time weld seam extraction method. The proposed framework enhances robustness through the sequential processing of historical frame data. First, an initial noise-free laser stripe image of the weld seam is acquired prior to arc ignition, from which the laser stripe region and slope characteristics are extracted. Subsequently, during welding, a dynamic region of interest (ROI) is generated for the current frame based on the preceding frame, effectively suppressing spatter and arc interference. Within the ROI, adaptive Otsu thresholding segmentation and morphological filtering are applied to isolate the laser stripe. An optimized RANSAC algorithm, incorporating slope constraints derived from historical frames, is then employed to achieve robust laser stripe fitting. The geometric center coordinates of the weld seam are derived through the rigorous analysis of the optimized laser stripe profile. Experimental results from various types of weld seam extraction validated the accuracy and real-time performance of the proposed method.

In the production process of high-voltage coils, a constant tension control system is designed to improve the quality of the transformer. The system is composed of a controller, execution structure, detection structure, etc. The active disturbance rejection control (ADRC), optimized by the Smith predictor (SP), is adopted to achieve constant tension control. The experiment results show that the tension control system based on the SP-ADRC has higher control accuracy, shorter stabilization time and stronger anti-interference ability compared with the traditional PID algorithm. The actual experiment shows that the constant tension control system of the high-voltage coil winding machine based on SP-ADRC has a superior control effect and high practical value.

To accurately estimate the 6D pose of objects, most methods employ a two-stage algorithm. While such two-stage algorithms achieve high accuracy, they are often slow. Additionally, many approaches utilize encoding–decoding to obtain the 6D pose, with many employing bilinear sampling for decoding. However, bilinear sampling tends to sacrifice the accuracy of precise features. In our research, we propose a novel solution that utilizes implicit representation as a bridge between discrete feature maps and continuous feature maps. We represent the feature map as a coordinate field, where each coordinate pair corresponds to a feature value. These feature values are then used to estimate feature maps of arbitrary scales, replacing upsampling for decoding. We apply the proposed implicit module to a bidirectional fusion feature pyramid network. Based on this implicit module, we propose three network branches: a class estimation branch, a bounding box estimation branch, and the final pose estimation branch. For this pose estimation branch, we propose a miniature dual-stream network, which estimates object surface features and complements the relationship between 2D and 3D. We represent the rotation component using the SVD (Singular Value Decomposition) representation method, resulting in a more accurate object pose. We achieved satisfactory experimental results on the widely used 6D pose estimation benchmark dataset Linemod. This innovative approach provides a more convenient solution for 6D object pose estimation.