Computationally Efficient IMU-Based Modeling of a Flexible Manipulator for Subsystem-Based Control
Year: 2025
Type: article
Abstract: Flexible manipulators are highly valued for their substantial power-to-weight ratio and effectiveness in harsh environments. However, tracking control of such systems presents significant challenges due to the existence of additional flexibility effects. To address this issue, this paper introduces a novel approach for the practical kinematic and dynamic modeling of a flexible structure as an open chain of rigid links utilizing feedback obtained from an inertial measurement unit (IMU) sensor network. As a result, the original partial differential equation (PDE) models are replaced by a set of approximated ordinary differential equations (ODE), which enables convenient real-time implementation in open-loop and closed-loop systems. Recovery of virtual system states based on IMU measurement is addressed using complimentary filtering, ensuring recovery of system states despite noise and drift effects. Subsequently, based on the approximated dynamic model of the open chain, a robust subsystem-based model predictive control (RSMPC) is proposed for the task-space control of the flexible manipulator. This controller is conveniently implementable in real time as it does not incorporate PDE calculations, and also forgoes the complexity of controlling a high degree-of-freedom (DoF) open-chain leveraging its subsystem-based design. The proposed method was experimentally verified using a long-reach flexible manipulator setup. Kinematic and dynamic estimated models are verified with respect to a ground truth laser-based position measurement system and load-cell force sensors. Furthermore, implementation of the RSMPC results in the improvement of the tracking precision as well as the generation of smaller and smoother inputs.
Institutions: Tampere University, University of Tampere
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $2,645
Funders: Academy of Finland, The Research Council
OpenAlex ID: W4413926010