Geometric feedback control of discrete-deposition SFF systems

Purpose - New applications of solid freeform fabrication (SFF) are arising, such as functional rapid prototyping and in situ fabrication, which push SFF to its limits in terms of geometrical fidelity due to the applications' inherent process uncertainties. Current closed-loop feedback control schemes monitor and manipulate SEE techniques at the process level, e.g. envelope temperature, feed rate. Closing the loop on the process level, instead of the overall part geometry level, leads to limitations in the types of errors that can be detected and corrected. The purpose of this paper is to propose a technique called greedy geometric feedback (GGF) control which closes the loop on the overall part geometry level. Design/methodology/approach - The overall part geometry is monitored throughout the print and, using a greedy algorithm, real-time decisions are made to serially determine the locations of subsequent droplets, i.e. overall part geometry is directly manipulated. A computer simulator and a physical experimental platform were developed to compare the performance of GGF to an open-loop control scheme. Root mean square surface height errors were measured under controlled uncertainties in droplet height, droplet radius of curvature, droplet positioning and mid-print part deformations. Findings - The GGF technique outperformed open-loop control under process uncertainties in droplet shape, droplet placement and mid-print part deformations. The disparity between performances is dependant on the nature and extent of the imposed process uncertainties. Practical implications - Future research will focus on improving the performance of GGF for specific cases by designing more complex greedy algorithmic scoring heuristics. Also, the technique will be generalized beyond heightmap representations of 3D spaces. Originality/value - The GGF technique is the first to close the loop on the overall part geometry level. GGF, therefore, can compensate for a broader range of errors than existing closed-loop feedback control schemes. Also, since the technique only requires the real-time update of a very limited set of heights, the technique is computationally inexpensive and widely applicable. By developing a closed-loop feedback scheme that addressed part geometry-level errors, SFF can be applied to more challenging in situ fabrication scenarios with less conventional materials.