Researchers at the Georgia Institute of Technology have developed this new method that can directly label a workspace for flexible manufacturing, classify images via supervised learning data sets, and detect bolts for bolt insertion. This makes the approach applicable to any workspace environment, even if it has difficult and variable conditions such as an aerospace environment.

The invention is a new, collaborative, and adaptive robot-based approach to complete drilling and fastening tasks autonomously in an adaptable, unstructured environments using an expert demonstrator and human operator. The human trains the robot to autonomously complete tasks by defining its environment and demonstrating how to locate, classify, and insert fasteners into a fuselage using a camera and 3D sensor that surveys the area. While the system starts with no information, it can use online and offline learning techniques to develop a data bank of information to utilize throughout the insertion process.