By uniting predictive modelling with robot driven experimentation, researchers accelerate the search for flexible conductors capable of withstanding extreme mechanical strain.
Researchers at the University of Maryland have developed an AI-integrated platform that automates the design of stretchable electrodes for wearable electronics and soft robotics. The platform combines robotics, machine learning and simulation tools to streamline the engineering of flexible electronic materials.
The proposed system is designed to reduce reliance on traditional trial-and-error experimentation, which can slow material development. By integrating automated experimentation with predictive modelling, the workflow aims to accelerate the identification of electrode designs that maintain electrical performance under significant mechanical strain. According to the researchers, the AI-enabled process produced electrodes capable of stretching to more than ten times their original length while retaining conductivity. Such properties are relevant for stretchable transistors, wearable and implantable sensors, deformable energy storage devices and soft robotic systems.
A key advantage of the approach lies in its ability to jointly address mechanical resilience and electrical functionality. Li’s group found that by controlling the micro texture of conductive layers, the material accommodates strain through geometric unfolding rather than fracturing. This mechanical response supports durability under repeated bending, twisting and stretching. The automation framework may also help researchers explore broader material design spaces more efficiently compared to conventional laboratory methods.
The AI-driven workflow integrates robot-automated experiments, neural network-based machine learning models and physics-based simulations. It uses a data-driven process to evaluate formulation, microstructure and processing parameters, enabling multi-objective optimization for mechanical and electrical performance targets.
Haochen Yang, the study’s first author and a doctoral student in chemical and biomolecular engineering, said, “This capability can help robots interact more safely and effectively with people and the physical world,” as team plans on further refining the platform and to enhance the AI and robotics integration with broader datasets and improved automation to speed up the identification of high-performing designs.
