AI-Driven Battery-Free Wireless Sensing of Hazardous Liquid Spills via a Frequency-Selective Surface in a Monostatic Antenna Configuration

Abstract

The detection of spills is paramount in safeguarding safety and mitigating environmental risks in sensitive environments, including laboratories and industrial facilities. Here, the novel artificial intelligence (AI)-driven, battery-free, and wireless sensing methodology are presented for detecting liquid spills using a monostatic wireless sensing system. The system consists of a frequency-selective surface (FSS) serving as the sensor, in conjunction with a horn antenna that functions as the readout equipment. The sensing structure features a $25 \times 25$ resonator array with a 7-mm periodicity, operating at a resonant frequency of 7.2 GHz. The system analyzes the renormalized $S_{11}$ response to quantify variations caused by the presence of liquid on the FSS, demonstrating a high sensitivity to isopropyl alcohol (IPA) spills. Using machine learning techniques, the framework generates $512 \times 512$ pixel masks delineating the affected area on the FSS, achieving an ${F}1$ -score exceeding 0.85 for spill localization. This sensing methodology shows potential for integration with augmented reality (AR) systems, enabling enhanced situational awareness and real-time spill localization. Future work aims to enhance the system’s capability to detect more hazardous materials and accurately classify them.