Research Analyzer
← Back ICRA 2026

Co-Design and Morphology-Guided Feedback Control: An Approach for Soft Robots

Nhan Huu Nguyen, Dinh Truong Do, Le Minh Nguyen, Van Ho

PDF

AI summary

Key figure (auto-extracted from paper)
Jointly optimizing soft robot morphology and closed-loop feedback control via simulation reliably yields adaptive, high-performing design-controller pairs.
Soft robotics Co-design Morphology control Reinforcement learning Proprioceptive feedback Simulation optimization

Problem

Soft robots struggle with complex control due to highly compliant bodies, vast design spaces, and a reliance on open-loop strategies that lack adaptive feedback.

Approach

A hierarchical simulation framework that co-evolves discrete structural parameters and continuous actuation variables alongside a deep reinforcement learning controller trained on proprioceptive strain feedback.

Key results

  • Reliable convergence to optimal design-control pairs in simulation
  • Successful integration of proprioceptive strain feedback for dynamic force regulation
  • Effective handling of heterogeneous design spaces via factorized discrete and continuous distributions
  • Demonstrated adaptive morphology enabling robust performance under varying contact conditions

Why it matters

Provides a scalable pathway to endow soft robots with morphological intelligence and closed-loop adaptability for complex, real-world tasks.

Abstract

Soft robots, with their highly compliant bodies, exhibit numerous unforeseen configurations that often defy engineering intuition and complicate control design. This work introduces a simulation-based co-optimization framework that jointly optimizes both morphology and control. Unlike exist- ing approaches that rely on oversimplified soft robot models or feed-forward controllers for simple tasks, our framework targets complex tasks that benefit from closed-loop feedback. The controller is trained over a hybrid design space combining discrete parameters, which define the nominal structure, and continuous parameters, which shift the morphology adaptively. The design distribution is iteratively manipulated to emphasize high-performing candidates until the optimal design–control pair emerges. Proprioceptive feedback in the form of me- chanical strain is integrated to provide the controller with awareness of morphological state and interaction dynamics. Demonstrations show that the framework converges reliably to optimal design–control solutions, validating the effectiveness of the proposed joint optimization strategy.

Index terms

Modeling Control and Learning for Soft Robots Evolutionary Robotics

Related papers