Skip to main navigation Skip to search Skip to main content

Dynamic inversion and optimal tracking control on the ball-plate system based on a linearized nonholonomic multibody model

  • A.G. Agúndez (Corresponding author-nrf)
  • , A. Saccon
  • , D. García-Vallejo
  • , E. Freire

Research output: Contribution to journalArticleAcademicpeer-review

90 Downloads (Pure)

Abstract

This paper addresses the optimal control of the ball-plate system, a well-known nonholonomic system in the context of nonprehensile manipulation, using a multibody dynamics approach. The trajectory tracking control of a steady-state circular motion of the ball on the plate, for any radius and potentially off-centric with respect to the plate's pivoting point, is achieved by designing a Linear-Quadratic Regulator. A spatial multibody model of the ball-plate system is considered. A key contribution is the analytical computation of the circular steady motion of the ball by dynamic inversion, including the control actions to achieve this reference solution. This enables the analytical computation of the linearized equations along this reference motion, resulting in a periodic linear time-varying (LTV) system, and the application of linear controllability criteria for LTV systems. A controllable linear system, involving the Cartesian coordinates of the contact point and the yaw angle of the sphere, is obtained using a convenient coordinate partition in the linearization. Compared to existing results on the same problem, closed-loop stability about the desired trajectory is achieved for any radius of the circular trajectory.

Original languageEnglish
Article number105795
Number of pages25
JournalMechanism and Machine Theory
Volume203
DOIs
Publication statusPublished - 15 Nov 2024

Keywords

  • Ball-plate system
  • Linearization
  • Multibody dynamics
  • Nonholonomic system
  • Trajectory tracking

Fingerprint

Dive into the research topics of 'Dynamic inversion and optimal tracking control on the ball-plate system based on a linearized nonholonomic multibody model'. Together they form a unique fingerprint.

Cite this