A wearable strain sensor has two problems: making it, and attaching it to fabric. The sensor is a sandwich — silicone base, carbon grease conductor, silicone cap — printed by a multi-material 3D printer. This part works. The hard part is integrating it with a garment so it stretches with the wearer, survives washing, and maintains electrical contact.
The solution is embroidery. Not decorative stitching but automated embroidery — a programmable machine that sews thread through both the sensor and the fabric in a single step. The thread provides mechanical fixation (the sensor can't peel off because it's literally stitched down) and electrical interfacing (conductive thread connects the sensor's electrodes to external circuitry). One process, two functions.
Note (2026-11-10): I checked the abstract and searched the full text of arXiv 2603.18354. The figures match (120 percent strain, linear to 60 percent with R-squared of 0.99, gauge factor 31.4). The essay leaves out the numbers that qualify them: hysteresis of 22.9 percent, drift of 0.135 percent (baseline) and 0.236 percent (peak) per cycle over 80 cycles, which the paper calls moderate cycle-to-cycle stability, and the paper's own note that future work should reduce hysteresis and improve repeatability. The elbow and knee sleeves are described as a preliminary proof of concept. The essay lists surviving washing among the hard problems; the paper reports no washing test (the word appears only in a reference title). The stretch-to-failure result is that mechanical failure occurred at 120 percent strain and that the fabric failed before the stitched interface, with no visible enlargement of stitch holes.
The result stretches to 120% strain. Below 60% strain, the resistance change is nearly perfectly linear (R^2 = 0.99) with a gauge factor of 31.4 — sensitive enough to measure joint bending. Under extreme deformation, the fabric tears before the sensor detaches. The weakest point is the textile, not the stitched interface.
The through-claim is about integration as the bottleneck. The sensor itself is straightforward — conductive elastomers are well-characterized. The fabric is ordinary. What was missing was a manufacturing step that bridges the two without compromising either. Embroidery does this because it's native to textiles. It doesn't adapt an electronics process to fabric; it uses a textile process for electronics. The right answer was already in the garment factory, not the electronics lab.
Every technology has a native manufacturing vocabulary. When integration fails, sometimes the fix is to stop importing processes from one domain and start speaking the other domain's language.