Data-driven pressure controller using proportional piezoelectric valves for soft pneumatic actuators

verfasst von

A. Pawluchin, T.-L. Habich, T. Seel, I. Boblan

Abstract

Pressure control forms the foundation for operating soft pneumatic actuators (SPAs). For effective motion or force control, however, the underlying pressure control must be both fast and accurate. This can be achieved either by placing the valve close to the actuator or by compensating for long pneumatic tubes through dynamic modeling. Tube compensation, however, is complex and difficult to implement, while direct valve mounting is often impractical because conventional proportional valves are large and heavy. To overcome these limitations, a compact, custom-designed 3/3-valve unit (CVU) based on Festo’s VEAE piezoelectric valves is developed. The CVU supports pressures up to 6bar, flow rates up to 70l/min and control bandwidths exceeding 9Hz. It is controlled using the presented data-driven approach, which eliminates the need for classical system identification and automatically adapts to different actuator volumes, resulting in high accuracy and simple deployment. The control scheme employs a two-stage, data-driven architecture based on single-shot Gaussian process (GP) regression. First, the inverse static flow characteristics of each valve are modeled, compensating for valve-to-valve variability without manual mass-flow identification. Second, the CVU is adapted to the actuator’s state-dependent volume, improving accuracy and robustness to external disturbances. In both stages, only the pressure derivative is used, avoiding the need for additional flow sensors or external test benches and keeping the approach lightweight and low-cost. The CVU with the data-driven control method was validated on an antagonistic pneumatic arm with pneumatic artificial muscles (PAMs) and benchmarked against a manually tuned PID controller, a feedback-linearized controller based on analytical system inversion and a commercially available VEAB valve unit. Across all tests, the CVU with GP-based control achieved highly accurate pressure tracking and disturbance rejection. All hardware (CAD) and development code (m-code) are released as open source.

Organisationseinheit(en)

Institut für Mechatronische Systeme

Typ

Artikel

Journal

MECHATRONICS

Band

113

Seiten

103428

Anzahl der Seiten

1

ISSN

0957-4158

Publikationsdatum

2026

Publikationsstatus

Veröffentlicht

Peer-reviewed

Ja

Elektronische Version(en)

https://doi.org/10.1016/j.mechatronics.2025.103428 (Zugang: Unbekannt)
https://www.sciencedirect.com/science/article/pii/S0957415825001370 (Zugang: Unbekannt)