Manipulation of Linear Deformable Objects with Non-Negligible Dynamics Grasped at Multiple Locations: A Stability-Guaranteed Approach

Daniel Feliu-Talegon, Cosimo Della Santina

Keywords: Robotic Soft Manipulation, Shape Regulation, Deformable Linear Objects, Actuation Coordinates.

TL;DR: Soft Robot Manipulation

Abstract:

Most existing research on deformable object manipulation
assumes the objects are lightweight and exhibit negligible
mechanical response, limiting the problem to quasi-static
conditions. However, many real-world objects—such as hoses,
pipes, and wiring harnesses—have significant dynamics that
must be considered during manipulation. This work addresses
that gap by proposing a closed-loop control framework that
explicitly models object dynamics and treats manipulation as
a shape-regulation problem. The system controls the object by
adjusting forces and torques at multiple fixed points along its
length. The approach is based on three main contributions: a
fully dynamic model for linear deformable objects using discrete
strain parameterization; an extension of actuation coordinates
to SE(3), enabling a structured yet naturally underactuated
control framework; and nonlinear feedback control laws with
clear conditions for convergence to desired steady-state shapes.
Simulation results on representative tasks show clear performance
improvements using the proposed model-based method.
Experimental validation with real-time closed-loop control and
online shape estimation further demonstrates the method’s
practical feasibility and effectiveness.

Submission Number: 37