Abstract: To ensure the final assembly of large aerospace cylindrical components is coordinated and interchangeable, the reserved machining allowance of the assembly interface of the large component is usually fettled before final assembly. Note that accurate positioning of the large component is critical to ensure the machining accuracy of the assembly interface. However, in the traditional method, the large component is manually marked and aligned, which has a negative impact on the machining quality and efficiency of the assembly surface. To cope with these problems, an in-situ pose measurement method for large cylinders based on binocular vision and prior data is proposed. In this method, the deep learning and edge detection vision algorithms are applied to identify and locate key features of the large cylinder, and the central position of the key feature can be located to a pixel in the original image of the large cylinder. Based on this, the spatial coordinates of the center position of the key feature can be solved using binocular vision 3D reconstruction method. Meanwhile, the actual pose of the large component is finally solved by combining the prior processing data of the large component and the 3D reconstruction results of the key features, which can improve the pose calculation accuracy. Finally, the in-situ pose measurement control system for the large cylinder is developed and validated using the proposed method, and the experimental results illustrate that the developed measurement control system is capable of efficiently realizing high-precision pose perception of the large cylinder in order to meet the production demand of in-situ finish machining of the assembly interface.

Key words: binocular vision, laser tracking measurement, prior processing data, large cylindrical component, pose measurement