Physics-guided neural controllers for compensating parasitic forces in high-precision mechatronics

Expanding markets for integrated circuits and 3D printing call for rapid development of a new generation of intelligent high-precision mechatronics, which can move mechanical stages with higher accuracy despite inherent parasitic forces. Therefore, this project will design a new type of data-driven intelligent controllers for compensating parasitic forces in high-precision mechatronics. The original idea is to develop physics-guided neural networks that are simpler to train and more robust compared to state-of-the-art deep neural networks. The resulting physics-guided neural controllers will be tested in an industrial linear motor for lithography machines with the aim of pushing accuracy from 100μm closer to 10μm in the presence of parasitic forces.