Experimental Closed-Loop Excitation of Nonlinear Normal Modes on an Elastic Industrial Robot

Filip Bjelonic, Arne Sachtler, Alin Albu-Schaffer, Cosimo Della Santina

Research output: Contribution to journalArticleScientificpeer-review

6 Citations (Scopus)
38 Downloads (Pure)

Abstract

Adding elastic elements to the mechanical structure should enable robots to perform efficient oscillatory tasks. Still, even characterizing natural oscillations in nonlinear systems is a challenge in itself, which nonlinear modal theory promises to solve. Therein eigenmanifolds generalize eigenspaces to mechanical systems with non-Euclidean metrics and thus characterize families of oscillations that are autonomous evolutions of the robot. Eigenmanifolds likewise provide a framework for deriving feedback controllers to excite and sustain these oscillations. Nevertheless, these results have been so far essentially theoretical. They have been applied on relatively low dimensional systems and almost exclusively in simulation. We aim to bridge the theory to the real-world gap with the present work and show that we can excite nonlinear modes in complex systems. To this end, we propose control strategies that can simultaneously stabilize numerically evaluated eigenmanifolds and sustain oscillations in the presence of dissipation. We then focus on the KUKA iiwa with simulated parallel springs as an example of the highly nonlinear and articulated system. We calculate all the nonlinear modes of the system, and we use the proposed strategies to excite the associated natural oscillations.

Original languageEnglish
Pages (from-to)1689-1696
JournalIEEE Robotics and Automation Letters
Volume7
Issue number2
DOIs
Publication statusPublished - 2022

Bibliographical note

Accepted Author Manuscript

Keywords

  • and Learning for Soft Robots
  • Control
  • Dynamics
  • Modeling
  • Motion Control

Fingerprint

Dive into the research topics of 'Experimental Closed-Loop Excitation of Nonlinear Normal Modes on an Elastic Industrial Robot'. Together they form a unique fingerprint.

Cite this