Machine-learned, waterproof MXene fiber-based glove platform for underwater interactivities

Waterproofness is of value and critical for electronic systems, especially involving underwater interactivities. Yet, the most reported works are related to waterproof flexible sensors. System-level waterproof platforms integrating with sensing, processing and wirelessly transmitting to provide consumer-grade performance have been seldom reported. Herein, we construct a machine-learning-assisted waterproof hybrid electronic system via the whole encapsulation technique integrating fiber sensors and processing center based on flexible printed circuit board (FPCB) so as to achieve fully underwater interactivities. The fiber sensors can be seamlessly integrated with the FPCB through knitting techniques, which exhibit excellent robustness, washability, and waterproofness, with resistance varying within in ± 0.6% in tape water, sweat, and saline solution for up to 15 days. The hybrid platform is demonstrated for an underwater piano playing “Happy birthday to you” and the remote control of a robotic hand in water using the multilayer perceptron that achieves a 98.1% classification accuracy for 20 kinds of hand gestures. Our work may promote advances in waterproof system-level electronics from device design, system integration to algorithm optimization.