Laser Scanning-Based Precision Defect Identification for Autonomous Robotic Part Repair${}^{*}$

In this article, a computer vision (CV)-based approach to precisely identify the defects of a part for autonomous robotic repair is proposed. Robotic autonomous repair is needed to retrofit high-value parts such as jet engine blades, as a robotic-based machining system offers advantages in manufacturing precision, low cost, flexibility, and adaptability. Although CV-based robotic manipulation has been explored for applications such as grasping and human–robot interaction, challenges emerge in CV-based robotic machining due to the more stringent precision and accuracy in part identification and quantification of the frame transformation, and identification of the defects on the part. The contribution of this work is the development of an experimental-based approach to precisely identify the frame transformation, and then a regressive method to identify and quantify the defects. We illustrate the performance of the proposed approach through experiments on a robotic autonomous repair example.