High accuracy keyway angle identification using VGG16-based learning method

Aligning perforated pipes in different manufacturing workstations is critical for ensuring the quality of the final product in the oil&gas industry. Keyways inside the pipes are typically used for alignment. In order to automate the alignment process using an industrial robot, the keyway angle must be identified accurately. Because the environmental conditions keep changing in the shop floor, current methods cannot satisfy the accurate alignment requirement. Recently, VGG16 has become one of the most effective ways to deal with vision-based problems with different lighting conditions. Therefore, this paper proposes a method based on the VGG-16 architecture to identify the keyway angle to satisfy the manufacturing requirement (angle accuracy ) in different lighting conditions. In order to demonstrate the effectiveness of the proposed method, two traditional methods, a commercial vision method, and a geometrical rule based method, are also investigated. The comparisons of the three methods show that the proposed method performs best, yielding an angle error less than 1 degrees in 98.38% of the testing images. The research results indicate that VGG-16 based method has significant potential to improve manufacturing processes.