In context
Soft robotics had emerged as a distinct field addressing the limitations of rigid robots in unstructured environments. By 2024, advances in materials and fabrication had produced remarkable prototypes, but precise control remained a bottleneck due to the inherent nonlinearity, hyper-redundancy, and large deformation of soft structures. This review systematically organized the fragmented control research to guide future development.
What was reported
The review, published in the journal Robot, categorized soft robot control technologies into three dichotomies: model-based versus model-free, open-loop versus closed-loop, and classical versus intelligent control. It examined the state of the art in modeling, feedback, and control algorithms, summarizing representative approaches and their trade-offs.
Modeling methods for fluid-driven soft robots include constant curvature (CC), piecewise constant curvature (PCC), variable curvature (VC), Lagrangian, Newton-Euler, beam theory, virtual rigid joint, and finite element methods. PCC is the most widely used kinematic approach, but it assumes constant curvature per segment, limiting accuracy. VC methods improve fidelity but increase computational load. Dynamic modeling often employs Lagrangian or Newton-Euler formulations to account for external loads and coupling effects.
For feedback, the review discussed sensor integration for closed-loop control, while intelligent control algorithms (e.g., adaptive, learning-based) were highlighted for handling model uncertainties. Key challenges include the lack of a unified modeling framework, high computational cost, and the difficulty of achieving precise position control under varying loads.
“How to precisely control a soft robot to achieve the desired grasping and manipulation has always been a widely concerned problem.”
Why it mattered
This review provided a structured roadmap for researchers and engineers, clarifying the strengths and limitations of current control strategies. By identifying gaps—such as the need for more accurate yet computationally efficient models and robust feedback methods—it helped direct future efforts toward making soft robots practical for industrial applications like delicate handling and adaptive manipulation in unstructured environments.
Source: 《机器人》期刊 (robot.sia.cn) · Published 2024-03-27 · “软体机器人建模与控制技术研究进展”