Multimaterials with different modulus can endow soft robots with embodied intelligence that deliver spatially varying deformation upon actuation. There is an increasing need for design tools that can rigorously and efficiently generate material layouts for desired motions. Here, we present a design paradigm for soft pneumatic multimaterial actuators by attaching a stiffer material layer as skeleton to softer inflated rubber, and develop a topology optimization based framework to automatically generate the skeleton layout that leads the actuator to achieve desired motions such as bending or twisting. Our method is enabled by a dynamic level set function to describe and track the topological change of the skeleton, large-deformation analysis compatible with the varying skeleton layout, and a gradient-based optimizer to govern the evolution of material layout, with the geometric and material nonlinearities taken into account. A forward geometric mapping and a backward design velocity mapping are constructed to allow manipulating the level sets on the planar space. We show that the design methodology is capable of generating high-performance bending and twisting actuators of cylindrical or customized cone shape. The simulation and experiment results show that, the bending actuator achieves a free bending angle 73∘^\circ and blocking force 2.05 N, and the twisting actuator achieves a large rotation angle of 143∘^\circ.

中文翻译：

---

用于所需运动的骨架增强软气动执行器的拓扑优化


具有不同模量的多种材料可以赋予软机器人具体智能，在驱动时产生空间变化的变形。人们越来越需要能够严格有效地生成所需运动的材料布局的设计工具。在这里，我们提出了一种软气动多材料执行器的设计范例，通过将较硬的材料层作为骨架附加到较软的充气橡胶上，并开发了基于拓扑优化的框架来自动生成骨架布局，使执行器实现所需的运动，例如弯曲或扭曲。我们的方法通过动态水平集函数来描述和跟踪骨架的拓扑变化、与变化的骨架布局兼容的大变形分析以及基于梯度的优化器来控制材料布局的演变，以及几何和考虑材料非线性。构造前向几何映射和后向设计速度映射以允许操纵平面空间上的水平集。我们表明，该设计方法能够生成圆柱形或定制圆锥形的高性能弯曲和扭转执行器。仿真和实验结果表明，弯曲驱动器实现了73∘^\circ的自由弯曲角度和2.05 N的阻挡力，扭转驱动器实现了143∘^\circ的大旋转角度。