Gaussian Processes for Advanced Motion Control

M.M. Poot (Corresponding author), Jim Portegies, Noud Mooren, Max van Haren, Max van Meer, Tom Oomen

Research output: Contribution to journal › Article › Academic › peer-review

13 Citations (Scopus)
170 Downloads (Pure)

Abstract

Machine learning techniques, including Gaussian processes (GPs), are expected to play a significant role in meeting speed, accuracy, and functionality requirements in future data-intensive mechatronic systems. This paper aims to reveal the potential of GPs for motion control applications. Successful applications of GPs for feedforward and learning control, including the identification and learning for noncausal feedforward, position-dependent snap feedforward, nonlinear feedforward, and GP-based spatial repetitive control, are outlined. Experimental results on various systems, including a desktop printer, wirebonder, and substrate carrier, confirmed that data-based learning using GPs can significantly improve the accuracy of mechatronic systems.
Original languageEnglish
Article number21011492
Pages (from-to)396-407
Number of pages12
JournalIEEJ Journal of Industry Applications
Volume11
Issue number3
DOIs
Publication statusPublished - 2022

Funding

Current research focuses on several aspects. First, Gaussian processes are being further deployed towards industrial applications. Second, fundamental developments include the use Gaussian processes in control techniques, including new repetitive control algorithms(72). Third, other techniques that are at the intersection of control, machine learning, and mechatronics are being developed, including parametric techniques such as physics-guided neural networks(73), sparse optimisation for subset selection in feedforward and learning(59), and reinforcement learning for model-free learning in mechatronics(74). Acknowledgment The authors would like to thank Gert Witvoet, Lennart Blanken, Robin van Es, Dragan Kostić, and former master students for their contributions to this research. This paper contains a survey of contributions from different applications. The authors gratefully acknowledge the following support. The contributions of Van Haren and Van Meer related to Sec. 5 and 6, respectively, were carried out when they were affiliated with ASM Pacific Technology. They are currently PhD candidates within the ECSEL Joint Undertaking under grant agreement 101007311 (IMOCO4.E). The Joint Undertaking receives support from the European Union’s Horizon 2020 research and innovation programme. The research of Mooren related to Sec. 7 was carried out within the IMOCO4.E consortium and received funding from the European Union H2020 program under grant agreement 637095 (FourByThree), and ECSEL-2016-1 under grant agreement 737453 (I-MECH). The research of Poot related to Sec. 4.3 is supported by ASM Pacific Technology. Moreover, the work in this paper is part of research programme VIDI with project number 15698, which is (partly) financed by the Netherlands Organisation for Scientific Research (NWO).

FundersFunder number
ASM Pacific Technology15698
European Union's Horizon 2020 - Research and Innovation Framework Programme
European Commission737453, ECSEL-2016-1, 637095
Nederlandse Organisatie voor Wetenschappelijk Onderzoek

    Keywords

    • feedforward control
    • gaussian processes
    • learning control

    Access to Document

    • pdfFinal published version, 2.46 MB

    Other files and links

      Fingerprint

      Dive into the research topics of 'Gaussian Processes for Advanced Motion Control'. Together they form a unique fingerprint.
      View full fingerprint

      Cite this