MEKAWIRE
ExpertarXiv

AI-guided robotic in situ fabrication of curvature- and tissue-adaptive wireless bioelectronics

Hiroyuki Tetsuka, Jiaju Ma, Naomi Kumano, Minoru Hirano
In 30 seconds

This preprint presents a robotic system that prints customized wireless electrotherapy electronics onto transparent wound dressings, using surface curvature and tissue properties to guide their electrical design. On a tissue-equivalent forearm phantom, measured resonance was 25.68 MHz against a simulated 25.16 MHz and a 25 MHz target. The work connects anatomical sensing to functional device design and fabrication, but does not establish wound-healing efficacy.

Research question

Can measured wound geometry, surface curvature and tissue dielectric properties jointly guide robotic fabrication of wireless wound electronics that retain their intended resonant behavior on curved surfaces?

Problem

A wireless device changes electrically when bent onto the body: curvature alters its resonant characteristics, while tissue adds dielectric and conductive loading. Printing a predefined pattern along a curved surface therefore does not ensure that the finished device operates at its intended frequency.

Previous approach

Vision-guided robotic printing generally adapts deposition paths to surface geometry. CARE customized electrodes and receiving coils using wound dimensions and considered dielectric loading, but fabricated patches before attachment and did not explicitly use measured local curvature in RF optimization. Manual transfer of sensing results into design was another automation gap.

New approach

An Epson C8-A701S six-degree-of-freedom arm, Orbbec Femto Bolt RGB-D camera and direct-ink-writing head form the platform. YOLO-seg identifies wounds; depth data supply dimensions and local curvature. Python automatically passes these measurements to MATLAB optimization of coil and interdigitated-electrode geometry, accounting for dressing and tissue loading and an approximate quality-factor metric. The design becomes surface-following paths and tool orientations. A receiving coil, insulating layer and electrodes are printed onto an already-applied polyurethane dressing. Conductive ink solids contain 85 wt% silver flakes, 5 wt% silver nanoparticles and 10 wt% PEO.

Results

Five wound phantoms, measured three times each, yielded width/length MAEs of 0.45/0.39 mm; area MAPE was 10.74% across 15 measurements, with systematic underestimation. Curvature MAE was 2.57 mm and MAPE 6.16% across nine measurements at nominal radii of 30, 40 and 50 mm. YOLO-seg detected 16/16 wounds under ambient lighting and each of three controlled color temperatures; color segmentation detected 9/16 under ambient lighting and 2/16 under each controlled condition. Forearm-phantom simulations gave 23.87 MHz planar and 25.16 MHz curved under tissue-equivalent loading, a 1.29 MHz shift (5.40%). Measured resonance was 25.68 MHz, differing from the curved simulation by 0.52 MHz (2.07%); the target was 25 MHz. The workflow took approximately 675 s: computation 65 s, four-pass printing 190 s and drying 420 s. Printed conductors maintained continuity through 1000 cycles at 50% compressive deformation, although resistance increased. Fabrication and RF validation used physical phantoms, not living subjects.

Limitations

No living-tissue evaluation establishes biological response, safety or therapeutic efficacy. The fabrication process does not correct target motion or deformation after scanning. RF models simplify tissue structure and omit effects requiring further study, including wound fluid, dressing hydration and spatial dielectric variation. Wound-area underestimation, curvature uncertainty and long-term durability under moisture and body motion remain concerns. Dressing placement and ink loading are manual, and the robotic infrastructure adds complexity. RF validation sample count and code availability: Not reported in the paper.

Industry impact

Wound-dressing manufacturers, bioelectronics developers and medical robotics integrators could use the workflow now for experimental customization on anatomically complex surfaces. Clinical products would first require evidence of safety and healing benefit, suitable stimulation settings, motion compensation and durability in wet wound environments. A timeline for clinical or commercial adoption is not reported in the paper.

The full text is not republished here because the paper's license does not allow it. Read the original on arXiv.