Gijs van Houtum, Graduate student, Eindhoven University of Technology
This presentation will detail a vision system, consisting of a camera and a line laser, that is able to detect the deposition in a straight line, but complex trajectories are hard due to the geometric configuration of the system. A model of the deposition is synthesized through system identification applied on data from straight line deposition. A sub-optimal input pressure policy can be obtained for the upcoming layer through the application of Model Predictive Control.
Feature size detection takes place after each deposition layer. Line scanning of the complete layer exposes feature sizes at all locations. Then the errors are used to calculate the input policy of the next layer. This approach is referred to as dynamic layer intermittent error detection and compensation.
A typical usage would be to apply straight line system identification at the start of the manufacturing to overcome changing dynamics through external influences like temperature and/or using different materials. The model is then used for the rest of the process as an estimator for layer corrective action deposition. This approach will be demonstrated in the context of fabricating biomedical prostheses using medical-grade silicone.