Search Results

With typically over 2.3 million parts, attached with over 3 million fasteners, it may be surprising to learn that approximately two out of every three fasteners on a twin aisle aircraft are fastened by hand. In addition the fasteners are often installed in locations designed for strength and not necessarily ergonomics. These facts lead to vast opportunities to automate this tedious and repetitive task. The solution outlined in this paper utilizes the latest machine vision and robotics techniques to solve these unique challenges. Stereo machine vision techniques find the fastener on the interior of an aerospace structure and calculate the 6DOF (Degrees of Freedom) location in less than 500ms. Once the fastener is located, sealed, and inspected for bead width and gaps, a nut or collar is then installed. Force feedback capabilities of a collaborative robot are used to prevent part damage and ensure the nut or collar are properly located on the fastener.

Inspection of fasteners prior to installation is critical to the quality of aerospace parts. Fasteners must be inspected for length/grip and diameter at a minimum. Inspecting the fasteners mechanically just prior to insertion can cause additional cycle time loss if inspection cannot be performed at the same time as other operations. To decrease fastener inspection times and to ensure fastener cartridges contain the expected fastener a system was devised to measure the fastener as it travels down the fastener feed tube. This process could be adapted to inspection of fasteners being fed to the process head of a running machine eliminating the mechanical inspection requirement and thus decreasing cycle time.

Accurate measurement of countersinks in curved parts has always been a challenge. The countersink reference is defined relative to the panel surface which includes some degree of curvature. This curvature thus makes accurate measurements very difficult using both contact and 2D non-contact measurements. By utilizing structured light 3D vision technologies, the ability to very accurately measure a countersink to small tolerances can be achieved. By knowing the pose of the camera and projector, triangulation can be used to calculate the distance to thousands of points on the panel and countersink surface. The plane of the panel is then calculated using Random Sample Consensus (RANSAC) method from the dataset of points which can be adjusted to account for panel curvatures. The countersink is then found using a similar RANSAC method.

In an attempt to be more flexible and cost effective, Aerospace Manufacturers have increasingly chosen to adapt a manufacturing style which borrows heavily from the Automotive industry. To facilitate this change in methodologies a system for locating robots has been developed which utilizes cameras for both locating and guidance of a mobile platform for a robot with drilling and fastening end effector.

Accurately measuring the length of a pintail type fastener is limited by the process of forming the fastener. When the pintail is formed its overall length is not dimensionally controlled. To accurately measure the grip of the bolt a vision system is utilized that finds the notch between the tail and bolt shank. The grip, diameter, and angle of the bolt prior to insertion are then measured. This method proves to be more accurate than measuring the bolt mechanically and provides a number of other advantages including; decreased measurement time, improved accuracy, FOD detection, and angle of the bolt in the fingers prior to insertion.