Application of template matching algorithm in welding robots

One of the key technologies for intelligent welding is opinion-based seam tracking and determination of the welding shape. The biggest challenge in robotic applications is the interaction with a dynamic environment limited by the capacity of visual sensors and the calculation cost of signal processing algorithms. The goal is to speed up the welding process and reduce costs. Welding process is accelerated by using image sensors and various image processing algorithms instead of operators to increase the processing speed. However, the adaptability and robustness of most image processing algorithms during welding applications is insufficient. In this study, an event-driven concrete system for feature extraction and object recognition has been designed as a new and efficient sensory approach in robotic applications. The matching event feature has been developed by using additional information for object recognition. The aim of this study is to determine which welding type is closer to the welding process performed by the robot. Thus, parameter selections and processing of results are facilitated according to the welding type, and whether the welding is done in a high-quality manner is checked. Test results show that the algorithm provides good adaptability for multiple typical welding seams.