6-D pose estimation method for large gear structure assembly using monocular vision

Abstract Measuring the pose of large gear structures during the assembly process using large volume measurement systems (LVMSs) is difficult owing to the shielding between the output and input shafts. In addition, collision damage caused by assembly errors occurs during the assembly process, which affects the assembly accuracy and product reliability. In this study, a 6-D pose estimation method based on the monocular vision was proposed, which comprises a feature detection method and a pose calculation method, which is based on key features and is improved by using the weighting coefficient that was applied to the assembly of large gear structures. Finally, the assembly system of large gear structures based on the monocular vision was built. The system shows that this method has high precision and efficiency and can be used in the automatic assembly of large gear structures instead of traditional manual methods.