Max Bartholdt

Max Bartholdt

Auto-generated profile, based on publicly available information

Max Bartholdt

Institute of Mechatronic Systems Leibniz University Hannover 30823 Garbsen Germany

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
25 January 2022 in IEEE Robotics and Automation Letters
Verified
Journal Article
Closed access
SESimon R. Eugster
Simon R. Eugster
JHJonas Harsch
Jonas Harsch
MBMax Bartholdt
Max Bartholdt
MHMarco Herrmann
Marco Herrmann
MWMats Wiese
Mats Wiese
GCGiuseppe Capobianco
Giuseppe Capobianco
To describe the mechanical behavior of a soft pneumatic actuator, we present a nonlinear rod theory, which recognizes the chamber pressurization as contribution to the resultant contact forces and couples. Accordingly, the pressure actuation can be considered as part of the rod’s constitutive laws, which relate the contact interactions with the rod’s kinematics and the chamber pressures. The theory allows for nonlinear constitutive laws, which are capable of describing strengthening or softening behavior of largely deformable materials. The theory was applied to a three-chamber actuator to predict its centerline as well as the cross-section orientations in static equilibrium. The resulting governing equations were spatially discretized using beam finite elements. We evaluated four different actuator models that are contained in the presented theory regarding describability and predictability of the end effector position. Existing rod models with linear elastic material laws can describe the tip-displacement with an averaged deviation of 9.1 mm in the training set and 11.4 mm in the test set. With a deviation of 2.8 mm and 4.9 mm in the training and test set, respectively, the most enhanced model, for which the pressure chamber radii increase with increasing pressure, is more than 2 times more accurate.

28 August 2024 in arXiv
Verified
Preprint
Open access
PDF file
HKHenrik Krauss
Henrik Krauss
THTim-Lukas Habich
Tim-Lukas Habich
MBMax Bartholdt
Max Bartholdt
TSThomas Seel
Thomas Seel
MSMoritz Schappler
Moritz Schappler
Physics-informed neural networks (PINNs) are trained using physical equations and can also incorporate unmodeled effects by learning from data. PINNs for control (PINCs) of dynamical systems are gaining interest due to their prediction speed compared to classical numerical integration methods for nonlinear state-space models, making them suitable for real-time control applications. We introduce the domain-decoupled physics-informed neural network (DD-PINN) to address current limitations of PINC in handling large and complex nonlinear dynamical systems. The time domain is decoupled from the feed-forward neural network to construct an Ansatz function, allowing for calculation of gradients in closed form. This approach significantly reduces training times, especially for large dynamical systems, compared to PINC, which relies on graph-based automatic differentiation. Additionally, the DD-PINN inherently fulfills the initial condition and supports higher-order excitation inputs, simplifying the training process and enabling improved prediction accuracy. Validation on three systems - a nonlinear mass-spring-damper, a five-mass-chain, and a two-link robot - demonstrates that the DD-PINN achieves significantly shorter training times. In cases where the PINC's prediction diverges, the DD-PINN's prediction remains stable and accurate due to higher physics loss reduction or use of a higher-order excitation input. The DD-PINN allows for fast and accurate learning of large dynamical systems previously out of reach for the PINC.