Abstract
Infant falls during critical developmental stages present severe injury risks due to immature motor control and undetectable trauma hazards. Existing monitoring solutions face limitations in detection accuracy, real-time response, and wearability, necessitating reliable monitoring solutions. This study introduces a wearable triboelectric sensor array (WTSA) for infant collision detection. The device integrates a flexible 3 × 3 array of polydimethylsiloxane (PDMS) hemispherical protrusions mounted on an Ecoflex substrate with copper electrodes. Structural deformation during impact generates triboelectric signals through contact electrification and electrostatic induction. The WTSA demonstrated high sensitivity, with voltage output scaling from -2 V at 0.5 g to -17 V at 3 g and exceptional stability during extended operation exceeding 48 h. It serves dual functions as an accelerometer and self-powered sensor, achieving approximately 42.5 nW peak power output and successfully charging capacitors to 2.8 V. Crucially, impact location identification exceeded 93.6% accuracy using a convolutional neural network (CNN) trained on voltage waveforms from all nine sensor units. Its flexible design ensures comfortable, nondisruptive wear during infant activities. By enabling real-time, precise assessment of collision dynamics and injury risk through machine learning, the WTSA provides a robust, self-powered solution for enhancing infant safety during critical developmental stages.