Data-Enabled Control of Low-Temperature Plasma Interactions with Complex Interfaces

Despite recent advances in learning-based and predictive control of low-temperature plasma processes, there remain major challenges towards effective control of plasma processes [1]. In particular, an important challenge arises from the need to adapt control policies after each process run using (often limited) observations of plasma-induced effects on a target interface that can only be measured between process runs. Control policy adaptation is necessary to account for variable characteristics of plasma and target surfaces across different subjects and treatment scenarios [2]. To this end, this talk presents a data-efficient, “globally” optimal strategy to adapt deep learning-based controllers that can be readily embedded on resource-limited hardware for portable medical devices. The proposed strategy employs multi-objective Bayesian optimization to adapt parameters of a deep neural network (DNN)-based control law using observations of closed-loop performance measures. The proposed strategy for adaptive DNN-based control is demonstrated experimentally on a cold atmospheric plasma jet with prototypical applications in processing of (bio)materials. 

[1] Anirudh, R., Archibald, R., Asif, M. S., Becker, M. M., Benkadda, S., Bremer, P. T., ... & Zhang, X. (2023). 2022 review of data-driven plasma science. IEEE Transactions on Plasma Science.

[2] Chan, K. J., Makrygiorgos, G., & Mesbah, A. (2023). Towards personalized plasma medicine via data-efficient adaptation of fast deep learning-based MPC policies. In 2023 American Control Conference (pp. 2769-2775).