SimTO: A simulation-based topology optimization framework for bespoke soft robotic grippers

TL;DR

SimTO addresses the challenge of designing bespoke soft robotic grippers for feature-rich objects by replacing manually prescribed loads with automatically extracted design-dependent contact loads from dynamic grasp simulations. It integrates a dynamic grasp simulator (Neo-Hookean solids) with a 2D SIMP-based topology optimization under a $f(\boldsymbol{\rho})$ objective that aligns deformation with gripper-object contact directions, enabling high-resolution morphology that is attuned to object features. The framework iterates between load extraction and topology optimization, yielding object-specific grippers (thousands of designs across multiple objects) that generalize to unseen objects and reveal a spectrum of grasping strategies. These results demonstrate the potential of automated, simulation-driven co-design of soft grippers, with implications for automated manufacturing, agriculture, and healthcare where delicate, complex objects must be handled safely and securely.

Abstract

Soft robotic grippers are essential for grasping delicate, geometrically complex objects in manufacturing, healthcare and agriculture. However, existing grippers struggle to grasp feature-rich objects with high topological variability, including gears with sharp tooth profiles on automotive assembly lines, corals with fragile protrusions, or vegetables with irregular branching structures like broccoli. Unlike simple geometric primitives such as cubes or spheres, feature-rich objects lack a clear "optimal" contact surface, making them both difficult to grasp and susceptible to damage when grasped by existing gripper designs. Safe handling of such objects therefore requires specialized soft grippers whose morphology is tailored to the object's features. Topology optimization offers a promising approach for producing specialized grippers, but its utility is limited by the requirement for pre-defined load cases. For soft grippers interacting with feature-rich objects, these loads arise from hundreds of unpredictable gripper-object contact forces during grasping and are unknown a priori. To address this problem, we introduce SimTO, a framework that enables high-resolution topology optimization by automatically extracting load cases from a contact-based physics simulator, eliminating the need for manual load specification. Given an arbitrary feature-rich object, SimTO produces highly customized soft grippers with fine-grained morphological features tailored to the object geometry. Numerical results show our designs are not only highly specialized to feature-rich objects, but also generalize to unseen objects.

Figures (8)

• Figure 1: The SimTO framework.Left: The inputs to SimTO include (i) a deformable, feature-rich object and (ii) a soft gripper whose dynamic grasping behaviour can be simulated. In this work, we used a soft gripper design scheme inspired by liu2018_TObenchmark_v2, whose end-effectors are soft fingers actuated by the compression of a sliding stage. Right: Given an arbitrary feature-rich object, SimTO generates bespoke soft fingers which conform to that object's shape under actuation.
• Figure 2: The SimTO method. This figure shows two iterations of SimTO. Each iteration consists of: (i) a dynamic simulation of a 3D compliant mechanism in the deformed frame $\mathcal{D}$; (ii) contact force extraction, where the simulated 3D forces are rotated into $\mathcal{R}$ and projected onto the $x\!-\!y$ plane; and (iii) topology optimization of the undeformed 2D design using these in-plane forces. The resulting 2D design is then extruded to initialize the next iteration.
• Figure 3: Our design domain is based on that of liu2018_TObenchmark_v2, but with three key differences. First, whereas they manually prescribed two dummy loads that encouraged designs to deform solely along the negative $y$-axis, ours deform along $N_f$ simulated gripper-object contact force directions. Second, we rotated the input force $\mathbf{f}_{in}$ by an additional $10^{\circ}$. Third, we did not enforce material retention along the lower $80\mathrm{mm}$ edge of the design domain.
• Figure 4: Numerical results of SimTO optimization runs for the five feature-rich objects shown in Fig. \ref{['fig:fig1']}. Each plot compares the geometric diversity and object lift time of all grippers. Designs on the resulting Pareto front are indicated by enlarged markers with black outlines.
• Figure 5: Ten soft designs. All designs were optimized with $E_g = 11.51 \,\text{MPa}$ and evaluated with $E_g$ and $E_o$ also equal to $11.51\,\text{MPa}$ (see Table \ref{['tab1']} for results). In this figure, each design is compressed by $d_c = 80\, \text{mm}$.