Robotic Non-Destructive Inspection of Aircraft. Phase 1.

摘要

Non-Destructive Inspection (NDI) of aircraft is known to be a time consuming, boring, and an error prone task for human inspectors. A method is needed for looking (inspecting) at an aircraft without reliability problems due to boredom while minimizing aircraft total inspection time during the life of the aircraft. The application of robotics to the Non-destructive Inspection (NDI) of aircraft was investigated. It was found that it is feasible to transform the Navy's robotic deriveter to a neural network based NDI robot. A feedforward multilayer neural network was found to be very reliable at detecting cracks around rivets. In addition, an efficient new manipulator path planning method using neural networks was found to be useful for the robotic aircraft NDI.