design soft robots

This work addresses the limitations of purely rigid robots in environmental adaptability and fully soft robots in load-bearing capacity and scalability by proposing an automated co-design framework for hybrid soft-rigid robots. For the first time in this domain, differentiable simulation coupled with gradient-based optimization is employed to jointly optimize free-form soft structures, rigid truss layouts, and multi-channel actuation configurations. The developed differentiable simulator integrates the Material Point Method (MPM) with Extended Position-Based Dynamics (XPBD), enabling end-to-end gradient optimization of coupled soft-rigid systems and automatically generating truss skeletons that efficiently transmit actuation forces and elicit effective gaits. Physical prototypes successfully reproduce the optimized locomotion patterns, and modal analysis confirms alignment between structural deformation modes and actuation frequencies, significantly enhancing locomotion performance.

This study addresses the challenge of systematically evaluating soft pneumatic actuators, whose performance is influenced by diverse structural factors yet lacks a unified framework for cross-design comparison in specific applications. The authors propose a classification scheme based on motion types—linear, bending, twisting, and omnidirectional—and establish clear relationships between structural features (such as braid angle, crease geometry, fiber orientation, chamber arrangement, asymmetry, and constraint layers) and actuator performance metrics. By clarifying the essential conditions for meaningful inter-study comparisons, this work reveals significant differences in pneumatic requirements and practicality among actuators that produce similar motion outputs. The framework provides valuable theoretical guidance for selecting and designing soft actuators tailored to applications in wearable devices, biomedical systems, and mobile robotics.

Traditional heuristic approaches struggle to precisely control the nonlinear mechanical behavior of pneumatic soft actuators to achieve desired deformations. To address this challenge, this work proposes the first gradient-based inverse design framework that integrates nonlinear finite element modeling, three-dimensional shape parameterization, and pneumatic actuation mechanics. By leveraging gradient-based optimization, the method directly tailors the actuator’s geometric configuration to realize complex, target deformation patterns. This approach overcomes the limitations of conventional design strategies, enabling high-fidelity customization of soft actuator behavior. Experimental validation demonstrates excellent agreement between simulated and measured deformations of the designed actuators, significantly enhancing the accuracy and capability of demand-driven soft actuator design.

This paper addresses the lack of systematic definition and analysis of adaptability in soft robotics. It first rigorously categorizes adaptability into *external adaptability*—encompassing responses to environmental conditions, object properties, geometric variations, and task dynamics—and *internal adaptability*, covering tolerance to manufacturing imperfections, material aging, and cross-platform control generalization. Through a comprehensive review of representative applications—including surgical, wearable, locomotive, and manipulative systems—the work systematically analyzes the synergistic interplay among structural design, soft sensing, and adaptive control. A unified analytical framework is proposed to address application-driven challenges, integrating materials science and robotics methodologies. The study identifies fundamental limitations in modeling fidelity, real-time soft sensing, and general-purpose adaptive control. These findings provide both theoretical foundations and practical pathways for enhancing the robustness and deployability of soft robots across diverse operational scenarios.

This study addresses the challenge of designing three-dimensional pneumatic soft actuators capable of efficient bending performance under large deformations. The authors propose a novel 3D nonlinear topology optimization framework based on a porous hyperelastic constitutive model, which, for the first time, enables topology optimization to handle extreme deformations in soft actuator design. The method consistently accounts for both geometric and material nonlinearities while incorporating manufacturability constraints. Leveraging stereolithography-based 3D printing, numerical simulations, and experimental validation, two optimized actuators were fabricated and demonstrated significant bending responses under prescribed pneumatic pressure. The close agreement between experimental results and simulation predictions confirms the effectiveness and advancement of the proposed approach.

This study addresses the absence of systematic design frameworks for large-scale, high-force soft robots under gravitational loads by proposing a geometry optimization method incorporating buckling constraints. By formulating a constrained optimization model and deriving closed-form analytical solutions, this approach enables both blocking force maximization and a priori performance evaluation. Physical experiments across three configurations validate that the method accurately predicts constraint satisfaction and maximum end-effector output force. This work establishes the first explicit analytical solution and design assessment framework for geometric optimization of large-scale soft manipulators, providing reliable theoretical tools for developing large-scale interactive soft robotic systems.

In soft robot design, shape, material distribution, and actuation are highly coupled, rendering traditional approaches inefficient for joint optimization due to the high computational cost of high-dimensional nonlinear simulations and the inapplicability of gradient-based methods. This work proposes a low-dimensional, structured design embedding based on shared basis functions that unifies these three aspects through a continuous deformation mapping and spatial material field encoding within a common latent space. The representational capacity of this approach predictably improves with the number of basis functions, remains compatible with black-box simulators, and enables end-to-end joint optimization. Experiments across multiple dynamic tasks demonstrate that the method achieves significantly better performance than neural network and voxel-based baselines using fewer parameters, and consistently outperforms sequential optimization strategies.

This study addresses the insufficient load-bearing stiffness of flexible textile systems by proposing a 3D braided shell structure integrated with active materials. A modular framework is constructed using assembled “braiding angle” units, while an eigenvalue analysis algorithm is employed to optimize the actuation and load-bearing mechanisms. This approach achieves high axial stiffness, low bending stiffness, and system-level damage resilience. Based on this methodology, five robotic prototypes were successfully developed, demonstrating stable locomotion under payloads up to 70 times their own weight while maintaining reliable performance even after extreme compression. This work establishes a novel structural design paradigm for high-payload soft robots.

This work addresses the limitations of conventional soft robotic arms in scalability, workspace coverage, and the balance between compliance and rigidity, which hinder their adaptability to diverse tasks. The authors propose a modular cable-driven soft robotic arm featuring a stacked multi-segment architecture that enables independent actuation and morphological reconfiguration, supporting on-demand extension. By integrating soft silicone materials, embedded tendon channels, and a dual-helix protective tendon design—combined with a tunable stiffness mechanism—the study achieves, for the first time, modular integration and independent control of multiple soft segments. Experimental results demonstrate that a three-segment configuration expands the planar workspace by 13-fold and increases volume by 38.9-fold compared to a single segment, while also revealing the influence of silicone hardness on compliance and load-bearing capacity.

To address insufficient modeling of pneumatic load dependency in soft pneumatic gripper design, this paper proposes a physics-informed topology optimization framework. It jointly models pneumatic loads using Darcy’s law and a drainage term, and employs a robust min-max strategy to concurrently optimize structural layout and erosion parameters. Leveraging Ogden hyperelastic constitutive modeling within finite element analysis and the Method of Moving Asymptotes (MMA), the approach first designs 2D compliant mechanisms and then extends them to 3D-printable, modular multi-arm architectures. The resulting gripper exhibits excellent controllable deformation across a wide pressure range and successfully manipulates diverse objects—irregularly shaped, variable-stiffness, and variable-weight—demonstrating strong adaptability and robust grasping performance.