Gaussian process-based nonlinearity compensation for pneumatic soft actuators
-
Alexander Pawluchin
,
Michael Meindl
Abstract
Highly compliant Pneumatic Soft Actuators (PSAs) have the potential to perform challenging tasks in a broad range of applications that require shape-adaptive capabilities. Achieving accurate tracking control for such actuators with complex geometries and material compositions typically involves many time-consuming and laborious engineering steps. In this work, we propose a data-driven learning-based control approach to address reference tracking tasks, incorporating self-adaptation in situ. We utilize a short interaction maneuver, recorded a priori, to collect the quasi-static data affected by severe hysteresis. Besides a linear feedback controller, we use two Gaussian process models to predict the feedforward control input to compensate for the nonlinearity in a one-shot learning setting. The proposed control approach demonstrates accurate tracking performance even under realistic varying configurations, such as alterations in mass and orientation, without any parameter tuning. Notably, training was achieved with only 25–50 s of experimental interaction, which emphasizes the plug-and-play capabilities in diverse real-world applications.
Zusammenfassung
Hochgradig nachgiebige pneumatische Aktuatoren (PSA) haben das Potenzial, anspruchsvolle Aufgaben in einer breiten Palette von Anwendungen zu bewältigen, insbesondere hinsichtlich ihrer Formanpassung und Adaptivität. Die präzise Folgeregelung solcher Aktuatoren mit komplexen Geometrien und Materialzusammensetzungen erfordert in der Regel viele zeitaufwändige und mühsame Entwicklungsschritte. In diesem Beitrag beschreiben wir einen datengetriebenen lernbasierten Regelungsansatz, um Folgeregelung und Selbstanpassung vor Ort zu integrieren. Wir nutzen ein kurzes Interaktionsmanöver, um quasi-statische Ein- und Ausgangsdaten des PSA zu generieren, die einer starken Hysterese unterliegen. Neben einem linearen Feedback-Regler verwenden wir eine Vorsteuerung mit zwei trainierten Gauß-Prozessen, um die Nichtlinearitäten in einem One-Shot-Lernverfahren zu kompensieren. Der vorgeschlagene Regelungsansatz zeigt ohne jegliche Parameteranpassungen eine präzise Folgeregelung, auch unter veränderlichen Konfigurationen, wie Änderungen der Masse und Ausrichtung. Bemerkenswert ist, dass das Training mit nur 25-50 s experimenteller Interaktion erreicht wurde, was die Plug-and-Play-Fähigkeiten in vielfältigen realen Anwendungen betont.
About the authors
Alexander Pawluchin received his B.Sc. degree in Numerical Simulations in 2015 and his M.Sc. degree in Mechatronics in 2019 from the Technical University Berlin, Germany. He is pursuing a doctoral degree jointly at Berlin University of Applied Sciences and Technical University Berlin. His research focuses on data-driven and learning-based control methods for pneumatic actuators and soft robots in general.
Michael Meindl received the M.Sc. degree in mechatronic engineering from the University of Applied Sciences Karlsruhe, Germany, in 2020. He is currently pursuing the PhD degree jointly at Leibniz University Hannover and FAU Erlangen-Nürnberg, and his research focuses on movement learning in robotic systems with the prime interest of combining methods from the fields of control theory and machine learning.
Ive Weygers is currently a postdoctoral researcher at FAU, Nürnberg, Germany with the Department of Artificial Intelligence in Biomedical Engineering (AIBE). He obtained a PhD in biomedical sciences within the Department of Rehabilitation sciences in 2021 and a Master in Engineering Electronics-ICT in 2017, both from KU Leuven, Leuven, Belgium. His research interests are on sensor fusion, inertial sensors, and autonomous systems.
Thomas Seel is the Director of the Institute of Mechatronic Systems at Leibniz Universität Hannover. He studied Engineering Cybernetics at OVGU Magdeburg and UC Santa Barabara and received the PhD in Control Engineering from TU Berlin in 2016. He has been a full professor at the Department AI in Biomedical Engineering of FAU Erlangen-Nürnberg since 2021 and has been appointed at LUH in 2023. His research interests include sensor fusion, dynamic learning, and sensorimotor AI in robotics and biomedical applications.
Ivo Boblan studied Electrical Engineering at the Technical University of Berlin (TU Berlin). He received his PhD degree from the Department of Bionik and Evolution Technique at the TU Berlin in 2009. Since 2016, he has been a full professor at the Berlin University of Applied Sciences (BHT) in the field of Humanoid Robotics and heads the Compliant Robotics Lab (CoRoLab) at the BHT. His research focuses on humanoid robotics and exoskeletons, especially in modeling and control of pneumatic systems.
Acknowledgment
We appreciate Festo SE & Co. KG for providing the hardware, with special thanks to Dr. W. Stoll for his support.
-
Research ethics: Not applicable.
-
Author contributions: The authors have accepted responsibility for the entire content of this manuscript and approved its submission.
-
Competing interests: The author state no conflict of interest.
-
Research funding: None declared.
-
Data availability: Not applicable.
References
[1] P. Polygerinos, et al., “Soft robotics: review of fluid‐driven intrinsically soft devices; manufacturing, sensing, control, and applications in human‐robot Interaction,” Adv. Eng. Mater., vol. 19, no. 1700016, 2017.10.1002/adem.201700016Search in Google Scholar
[2] H. Masashi, M. Takamitsu, N. Tomoyuki, T. Tatsuya, and M. Jun, “User-robot collaborative excitation for pam model identification in exoskeleton robots,” in 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2017, pp. 3063–3068.10.1109/IROS.2017.8206145Search in Google Scholar
[3] J. Burgner-Kahrs, D. C. Rucker, and H. Choset, “Continuum robots for medical applications: a survey,” IEEE Trans. Robot., vol. 31, no. 6, pp. 1261–1280, 2015. https://doi.org/10.1109/tro.2015.2489500.Search in Google Scholar
[4] J. Walker, et al.., “Soft robotics: a review of recent developments of pneumatic soft actuators,” Actuators, vol. 9, no. 1, p. 3, 2020. https://doi.org/10.3390/act9010003.Search in Google Scholar
[5] T. George Thuruthel, Y. Ansari, E. Falotico, and C. Laschi, “Control strategies for soft robotic manipulators: a survey,” Soft Robot., vol. 5, no. 2, pp. 149–163, 2018. https://doi.org/10.1089/soro.2017.0007.Search in Google Scholar PubMed
[6] C. Della Santina, C. Duriez, and D. Rus, “Model-based control of soft robots: a survey of the state of the art and open challenges,” IEEE Control Syst., vol. 43, no. 3, pp. 30–65, 2023. https://doi.org/10.1109/mcs.2023.3253419.Search in Google Scholar
[7] V. Falkenhahn, T. Mahl, A. Hildebrandt, R. Neumann, and O. Sawodny, “Dynamic modeling of bellows-actuated continuum robots using the euler–Lagrange formalism,” IEEE Trans. Robot., vol. 31, no. 6, pp. 1483–1496, 2015. https://doi.org/10.1109/tro.2015.2496826.Search in Google Scholar
[8] D. Mueller, A. Raisch, A. Hildebrandt, and O. Sawodny, “Nonlinear model based dynamic control of pneumatic driven quasi continuum manipulators,” in 2020 IEEE/SICE International Symposium on System Integration (SII), 2020, pp. 277–282.10.1109/SII46433.2020.9026206Search in Google Scholar
[9] F. Schreiber, et al.., “Tracking control with hysteresis compensation for manipulator segments driven by pneumatic artificial muscles,” in 2011 IEEE International Conference on Robotics and Biomimetics, 2011, pp. 2750–2755.10.1109/ROBIO.2011.6181721Search in Google Scholar
[10] X. Luo, M. Xiao, Y. Ding, and H. Ding, “Hysteresis modeling and compensation of a pneumatic end-effector based on Gaussian process regression,” Sens. Actuators, A, vol. 315, 2020, Art. no. 112227. https://doi.org/10.1016/j.sna.2020.112227.Search in Google Scholar
[11] D. Nguyen-Tuong and J. Peters, “Model learning for robot control: a survey,” Cogn. Process., vol. 12, no. 4, pp. 319–340, 2011. https://doi.org/10.1007/s10339-011-0404-1.Search in Google Scholar PubMed
[12] M. T. Gillespie, C. M. Best, E. C. Townsend, D. Wingate, and M. D. Killpack, “Learning nonlinear dynamic models of soft robots for model predictive control with neural networks,” in 2018 IEEE International Conference on Soft Robotics (RoboSoft), 2018, pp. 39–45.10.1109/ROBOSOFT.2018.8404894Search in Google Scholar
[13] P. Hyatt, D. Wingate, and M. D. Killpack, “Model-based control of soft actuators using learned non-linear discrete-time models,” Front. Robot. AI, vol. 6, 2019, Art. no. 22. https://doi.org/10.3389/frobt.2019.00022.Search in Google Scholar PubMed PubMed Central
[14] A. P. Sabelhaus and C. Majidi, “Gaussian process dynamics models for soft robots with shape memory actuators,” in 2021 IEEE 4th International Conference on Soft Robotics (RoboSoft), 2021, pp. 191–198.10.1109/RoboSoft51838.2021.9479294Search in Google Scholar
[15] C. Relaño, J. Muñoz, and C. A. Monje, “Gaussian process regression for forward and inverse kinematics of a soft robotic arm,” Eng. Appl. Artif. Intell., vol. 126, 2023, Art. no. 107174. https://doi.org/10.1016/j.engappai.2023.107174.Search in Google Scholar
[16] T.-L. Habich, S. Kleinjohann, and M. Schappler, “Learning-based position and stiffness feedforward control of antagonistic soft pneumatic actuators using Gaussian processes,” in 2023 IEEE International Conference on Soft Robotics (RoboSoft), 2023, pp. 1–7.10.1109/RoboSoft55895.2023.10122057Search in Google Scholar
[17] C. Della Santina, et al.., “Controlling soft robots: balancing feedback and feedforward elements,” IEEE Robot. Autom. Mag., vol. 24, no. 3, pp. 75–83, 2017. https://doi.org/10.1109/mra.2016.2636360.Search in Google Scholar
[18] M. Meindl, D. Lehmann, and T. Seel, “Bridging reinforcement learning and iterative learning control: autonomous motion learning for unknown, nonlinear dynamics,” Front. Robot. AI, vol. 9, 2022, Art. no. 793512. https://doi.org/10.3389/frobt.2022.793512.Search in Google Scholar PubMed PubMed Central
[19] C. E. Rasmussen and C. K. I. Williams, Gaussian Process for Machine Learning (Adaptive computation and Machine learning), 3. Print, London, England, The MIT Press, 2006.10.7551/mitpress/3206.001.0001Search in Google Scholar
[20] A. Pawluchin, M. Michael, T. Seel, and I. Boblan, “Accurate and agile control of a pneumatic robotic actuator by gp-based feedforward learning,” in Proceedings on Automation in Medical Engineering, 2023.Search in Google Scholar
© 2024 Walter de Gruyter GmbH, Berlin/Boston
Articles in the same Issue
- Frontmatter
- Editorial
- Special issue AUTOMED
- Survey
- Reviewing the potential of hearables for the assessment of bruxism
- Applications
- Model predictive control of blood glucose in critically ill patients using Gaussian processes
- Muscle fatigue detection based on sEMG signal using autocorrelation function and neural networks
- A switching lung mechanics model for detection of expiratory flow limitation
- Physiological hardware-in-the-loop test bench for mechanical ventilation
- Gaussian process-based nonlinearity compensation for pneumatic soft actuators
- Design, development, and optimization of the Somnomat Casa: a rocking bed for sleep studies and nocturnal interventions in home settings
- Tools
- Classroom-ready open-source educational exoskeleton for biomedical and control engineering
- An open-source research platform for mechanical ventilation based on Simulink® and STM32 nucleo
- Designing the user interface of a ventilator under the constraints of a pandemic
Articles in the same Issue
- Frontmatter
- Editorial
- Special issue AUTOMED
- Survey
- Reviewing the potential of hearables for the assessment of bruxism
- Applications
- Model predictive control of blood glucose in critically ill patients using Gaussian processes
- Muscle fatigue detection based on sEMG signal using autocorrelation function and neural networks
- A switching lung mechanics model for detection of expiratory flow limitation
- Physiological hardware-in-the-loop test bench for mechanical ventilation
- Gaussian process-based nonlinearity compensation for pneumatic soft actuators
- Design, development, and optimization of the Somnomat Casa: a rocking bed for sleep studies and nocturnal interventions in home settings
- Tools
- Classroom-ready open-source educational exoskeleton for biomedical and control engineering
- An open-source research platform for mechanical ventilation based on Simulink® and STM32 nucleo
- Designing the user interface of a ventilator under the constraints of a pandemic
