Model identification and vision-based H_∞ position control of 6-DoF cable-driven parallel robots

ABSTRACT

This paper presents methodologies for the identification and control of 6-degrees of freedom (6-DoF) cable-driven parallel robots (CDPRs). First a two-step identification methodology is proposed to accurately estimate the kinematic parameters independently and prior to the dynamic parameters of a physics-based model of CDPRs. Second, an original control scheme is developed, including a vision-based position controller tuned with the H_∞ methodology and a cable tension distribution algorithm. The position is controlled in the operational space, making use of the end-effector pose measured by a motion-tracking system. A four-block H_∞ design scheme with adjusted weighting filters ensures good trajectory tracking and disturbance rejection properties for the CDPR system, which is a nonlinear-coupled MIMO system with constrained states. The tension management algorithm generates control signals that maintain the cables under feasible tensions. The paper makes an extensive review of the available methods and presents an extension of one of them. The presented methodologies are evaluated by simulations and experimentally on a redundant 6-DoF INCA 6D CDPR with eight cables, equipped with a motion-tracking system.

KEYWORDS:

• Cable-driven parallel robots
• model identification
• vision-based control
• H_∞ control
• tension distribution

Acknowledgments

This work has been made possible by a PhD grant from the French Ministry of Education and Research, a funding 'Preciput ANR' from Strasbourg University and a state funding from the State-Region Project Contract (CPER).

Notes

1. The H_∞ norm of a linear system is the maximum singular value of its complex gain over frequency. It is also the maximum amplification when considering the ${\mathcal{L}}_2$ norm on signals. For a system G(s) of input u(t) and output y(t), ${\left|\right|G\left(s\right)\left|\right|}_{\infty }={max}_{\omega \in \mathbb{R}}\bar{\sigma }\left(G\left(j\omega \right)\right)={max}_{u\left(t\right)}\frac{{\left|\right|y\left(t\right)\left|\right|}_2}{{\left|\right|u\left(t\right)\left|\right|}_2}$.