Affordable High-Sensitivity Flexible Strain Sensor Integrated with Machine Learning for Tracking In-Air Handwriting and Leg Motion

Abstract

The designed flexible fiber strain sensor is fabricated by infiltrating superconductive carbon black into the SEBS substrate, exhibiting a small volume, high gauge factor (GF), wide strain range, excellent stability, and low cost. At a strain of 550%, GF ≈ 1909.5, while the strain detection range exceeds 620%. After 2000 cycles of tensile and compressive tests, the sensor maintains remarkable stability. In air-handwriting applications, the recognition of four commonly used characters was achieved based on changes in electrical signals. To further enhance the accuracy and efficiency of signal differentiation under large strain conditions, the multifusion machine learning algorithm was integrated into leg posture detection, achieving an accuracy of 0.9854. These results strongly support the advancement of flexible sensors in intelligent control, human–computer interaction, and motion monitoring.