Direct Data-Driven Predictive Control for a Three-dimensional Cable-Driven Soft Robotic Arm

TL;DR

Controlling 3D soft robots is challenging due to nonlinear, history-dependent dynamics. The authors apply a DeePC framework with SVD-based dimension reduction to a new cable-driven soft arm, using offline Hankel data to predict future behavior and a regularized optimization to track references. Experiments on fixed-point reaching and 3D trajectory tracking show that DeePC achieves higher accuracy and robustness than a baseline model-based controller, validating the approach for complex soft dynamics. The work offers a practical, data-driven control pathway and provides open-source hardware and code to foster further development in soft robotics.

Abstract

Soft robots offer significant advantages in safety and adaptability, yet achieving precise and dynamic control remains a major challenge due to their inherently complex and nonlinear dynamics. Recently, Data-enabled Predictive Control (DeePC) has emerged as a promising model-free approach that bypasses explicit system identification by directly leveraging input-output data. While DeePC has shown success in other domains, its application to soft robots remains underexplored, particularly for three-dimensional (3D) soft robotic systems. This paper addresses this gap by developing and experimentally validating an effective DeePC framework on a 3D, cable-driven soft arm. Specifically, we design and fabricate a soft robotic arm with a thick tubing backbone for stability, a dense silicone body with large cavities for strength and flexibility, and rigid endcaps for secure termination. Using this platform, we implement DeePC with singular value decomposition (SVD)-based dimension reduction for two key control tasks: fixed-point regulation and trajectory tracking in 3D space. Comparative experiments with a baseline model-based controller demonstrate DeePC's superior accuracy, robustness, and adaptability, highlighting its potential as a practical solution for dynamic control of soft robots.

Figures (6)

• Figure 1: Fabrication process of the soft robot: (a–c) sequential setup steps prior to silicone casting, and (d) the completed soft robot after curing. Each panel is labeled to indicate the corresponding components.
• Figure 2: Hardware setup of the soft robot, illustrating (a) the actuation system and (b) the low-level control system.
• Figure 3: Experimental setup with motion capture cameras and a test frame.
• Figure 4: Experimental results of DeePC control. (a–b) Tracking performance of bending angle $\phi_b$ and orientation angle $\gamma_g$. (c–d) Corresponding tracking errors.
• Figure 5: Trajectory tracking control results. (a) Comparison of 3D tracking performance. (b) Comparison of 2D tracking performance in the $x$–$y$ plane.