2025/09/15 by Eron Ristich, Ristich, Eron, Jiahe Wang +8
Computer Science · Engineering · Physics and Astronomy · #Advanced Mathematical Modeling in Engineering #Elasticity and Material Modeling #FOS: Computer and information sciences #Model Reduction and Neural Networks #Robotics (cs.RO)
paper · pdf · doi:10.48550/arxiv.2509.11567
openalex publication_date 2025/09/15 · openalex created_date 2025/10/12 · openalex updated_date 2026/07/28
Soft continuum robots can allow for biocompatible yet compliant motions, such as the ability of octopus arms to swim, crawl, and manipulate objects. However, current state-of-the-art continuum robots can only achieve real-time task-space control (i.e., tip control) but not whole-shape control, mainly due to the high computational cost from its infinite degrees of freedom. In this paper, we present a data-driven Koopman operator-based approach for the shape control of simulated multi-segment tendon-driven soft continuum robots with the Kirchhoff rod model. Using data collected from these simulated soft robots, we conduct a per-segment projection scheme on the state of the robots allowing for the identification of control-affine Koopman models that are an order of magnitude more accurate than without the projection scheme. Using these learned Koopman models, we use a linear model predictive control (MPC) to control the robots to a collection of target shapes of varying complexity. Our method realizes computationally efficient closed-loop control, and demonstrates the feasibility of real-time shape control for soft robots. We envision this work can pave the way for practical shape control of soft continuum robots.