i-Tac: Inverse Design of 3D-Printed Tactile Elastomers with Scalable and Tunable Optical and Mechanical Properties

TL;DR

i-Tac introduces an inverse design pipeline for creating 3D-printed tactile elastomers with customizable optical and mechanical properties, reducing design iterations and enabling scalable, targeted material fabrication.

Contribution

The paper presents a novel inverse design framework using multi-material additive manufacturing and response surface models to efficiently tailor elastomer properties for tactile sensors.

Findings

Validated the inverse design approach with experimental results

Achieved targeted transparency and hardness in fabricated elastomers

Reduced iterative design process compared to conventional methods

Abstract

Elastomers are central to vision-based tactile sensors (VBTSs), where they transduce external contact into observable deformation. Different VBTS architectures, however, require distinct optical and mechanical properties, particularly transparency and hardness. Conventional elastomer design relies on a forward, trial-and-error optimisation process from material preparation to property evaluation, which is inefficient and offers limited property scalability and target tunability. In this work, we present i-Tac, an inverse design pipeline for tailoring 3D-printed tactile elastomers with target optical and mechanical properties. Inspired by the composite structure of the human dermis, i-Tac exploits multi-material PolyJet additive manufacturing with three complementary resins. A mixture design methodology is employed to characterise the printed elastomers and establish response surface models (ReSMs) that map material compositions to functional properties, thereby defining a scalable property space. Based on user-defined targets, a desirability-function-based multi-objective optimisation is then performed to identify feasible composition regions and derive an optimal operating window for fabrication. This enables elastomers with desired properties to be manufactured in a single iteration, thereby achieving efficient target tunability. Experimental results validate the proposed i-Tac framework in terms of both property scalability and inverse design performance, showing that i-Tac can effectively tailor elastomer transparency and hardness while reducing the iterative burden of conventional forward design. By fabricating physical sensor samples from both commercial and custom designs, the proposed framework further demonstrates the potential of inverse-designed, monolithically manufactured elastomers for customisable VBTS fabrication.