Control of Laser Coating Removal Process, 10-R8385
Michael P. Rigney
Thomas G. Whitney III
Michael O. Blanton Jr.
Thomas E. Lyons Jr.
Inclusive Dates: 04/01/13 – 09/17/14
Background — Use of high-power lasers for coating (paint) removal from aircraft surfaces offers a significant improvement in process efficiency and reduction of both consumables and waste streams relative to alternate approaches (chemical stripping, media blasting). The ability of lasers to quickly ablate paint from surfaces comes with the challenge of controlling laser power to achieve a selective coating removal goal. A sensing and control system supporting selective coating removal allows for removing the top coat(s) while leaving the primer layer, overlapping processing passes, and minimizing energy input to the substrate. SwRI has long history in the design and deployment of large robotic systems employing media-blasting for coating removal from military fighter aircraft. This research, combined with our capabilities in precision mobile robot systems, supports developing coating removal systems for large military and commercial transport aircraft.
Approach — The objective of this research was to implement a coating state classifier supporting selective coating removal with a high-power laser. The sensing, classifier, and control systems must handle: 1) wide variation in coatings and substrates (thickness, color, laser ablation response), 2) subtle differences between some coatings and substrates, 3) observing a mixture of top coat and underlying material (coating or substrate) within the sensed incremental surface regions during the laser ablation process, 4) the presence of combustion products (flame, smoke, illumination) in the sensed region, and 5) a high process cycle rate (300 Hz).
Test panels representing a wide variety of commercial and military paint systems (primer and top-coat combination) were fabricated or procured. Regions of test panels were processed open-loop by the laser ablation scanner at numerous power levels (Figure 1). Regions were assigned a coating removal state indicating whether insufficient, excess, or the correct amount of coating had been removed. Image data acquired during and subsequent to open-loop processing was used to implement a coating removal state classifier based on features sensitive to image color, intensity gradients, and texture. Numerous features and classifiers were evaluated, followed by feature down-selection to achieve real-time assessment while retaining good classification performance. The output from the coating state classifier fed into a PID control loop to modulate laser power.
Accomplishments — Imaging and illumination system components were refined to obtain improved image quality. Synchronization between image acquisition and the laser scanner was modified to enable laser path measurement, supporting system operation on surfaces with variable curvature. Coating state classifiers provided an RMS error of 1 percent and 3 percent for classifiers trained to evaluate combinations of two- and eight-paint systems, respectively, and provided excellent performance in discriminating between observations representing five or more coating removal states. Image analysis algorithms and coating state classifiers operated in real-time (300 Hz) and successfully provided closed loop control of the laser ablation process (Figure 2).