| About this Abstract |
| Meeting |
MS&T23: Materials Science & Technology
|
| Symposium
|
Advances in Dielectric Materials and Electronic Devices
|
| Presentation Title |
Design for in-situ Computer Vision-based Automation of Drop-on-Demand Inkjet Drop Formation Optimization |
| Author(s) |
Maximilian Estrada, Ruyan Guo, Amar Bhalla |
| On-Site Speaker (Planned) |
Maximilian Estrada |
| Abstract Scope |
The process of drop optimization in a drop-on-demand inkjet printing process was automated using various neural network architectures. Currently, drop optimization is performed manually: an operator modifies voltage waveform parameters in response to real-time (stroboscopic) macro-lens camera footage of the printhead during jetting. The automation task was subdivided into three tasks: intrinsic representation, extrinsic representation, and property-relation representation. A Convolutional Recurrent Neural Network and a 3D Convolutional Neural Network were compared for intrinsic representation (interpreting drop quality from video); a fully connected network and an Long Short Term Memory (LSTM) will be compared for extrinsic representation (voltage waveform recommendation), and a Graph Neural Network and an LSTM will be compared for property-relation representation (relating the voltage waveform to drop quality). |