DropleX: Liquid sensing on tablet touchscreens

2025-11-04

AI Summary

This work introduces a non-intrusive liquid sensing paradigm leveraging off-the-shelf capacitive touchscreens to address two longstanding challenges: micro-liter-scale droplet identification and detection of contamination or adulteration within sealed containers. Methodologically, we uniquely disable the screen’s adaptive filtering mechanism and repurpose the physics of raindrop interference—formulating a physics-driven signal processing pipeline integrated with a learned classifier for compositional analysis. Crucially, no hardware modification is required; only raw capacitive data from standard touchscreens is utilized. Our approach achieves 96–99% accuracy in detecting micro-liter adulterants in carbonated beverages, wine, and milk; 93–96% accuracy in identifying trace-level compounds; and 86–96% accuracy in container-penetrating liquid detection. The core contribution lies in pioneering the repurposing of consumer-grade touchscreens as a low-cost, ubiquitous, highly sensitive, non-destructive, and easily deployable liquid sensing platform.

Technology Category

• Computer Vision: Object Detection & Categorization
• Humans and AI: Brain-Sensing and Analysis
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Security and Privacy: Data transparency and provenance
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

We present DropleX, the first system that enables liquid sensing using the capacitive touchscreen of commodity tablets. DropleX detects microliter-scale liquid samples, and performs non-invasive, through-container measurements to detect whether a drink has been spiked or if a sealed liquid has been contaminated. These capabilities are made possible by a physics-informed mechanism that disables the touchscreen's built-in adaptive filters, originally designed to reject the effects of liquid drops such as rain, without any hardware modifications. We model the touchscreen's sensing capabilities, limits, and non-idealities to inform the design of a signal processing and learning-based pipeline for liquid sensing. Our system achieves 96-99% accuracy in detecting microliter-scale adulteration in soda, wine, and milk, 93-96% accuracy in threshold detection of trace chemical concentrations, and 86-96% accuracy in through-container adulterant detection. Given the predominance of touchscreens, these exploratory results can open new opportunities for liquid sensing on everyday devices.

Problem

• Detects liquid adulteration through tablet touchscreens
• Enables non-invasive through-container contamination sensing
• Disables built-in filters for microliter-scale liquid analysis

Innovation

• Uses tablet touchscreen for liquid sensing
• Disables built-in filters via physics-informed mechanism
• Employs signal processing and learning pipeline