AI-Powered Sensor Ensuring Food Quality

The AI based sensor (artificial tongue) developed by Penn State University (PSU) which can instantly detect food freshness and safety with over 95% accuracy.

Researchers at Penn State University, Pennsylvania have developed an AI-driven electronics sensor capable of analysing subtle differences in liquids, significantly advancing food safety and diagnostic applications. This innovative device not only detects variations in liquid samples but also identifies potential safety issues in various foods, offering results within seconds. Designed with food safety professionals, quality assurance teams, and health diagnostics in mind, this technology provides a reliable, fast tool for assessing quality and safety standards across sectors. By allowing the AI to define its own assessment parameters, the device achieved impressive accuracy, highlighting its potential for broader applications in both food production and medical diagnostics.

The sensor uses a neural network and graphene-based ion-sensitive field-effect transistors, which allow it to assess different types of liquids, including milk, sodas, coffee, and fruit juices. In addition to detecting subtle variations such as water content in milk or spoilage indicators in juice the device can classify the authenticity and freshness of each sample. According to Saptarshi Das, lead researcher & professor of engineering, PSU, “We’re trying to make an artificial tongue, but the process of how we experience different foods involves more than just the tongue.”

– Advertisement –

At the core of this technology is a neural network designed to mimic the gustatory cortex, the brain region responsible for interpreting taste beyond simple sensory signals. This AI system, when trained on human-designated parameters, provided solid accuracy. However, accuracy increased dramatically to over 95% when the neural network used machine-derived criteria, revealing insights into the AI’s decision-making process. Andrew Pannone, doctoral student and co-author , PSU, explained, “We used a method called Shapley additive explanations, which allows us to ask the neural network what it was thinking after it makes a decision.”

This AI-powered sensor has applications across industries, serving food production, quality assurance, and health diagnostics by delivering quick, non-invasive assessments of product quality. By assessing the data holistically, the neural network adjusts to sample variations, much like how humans perceive differences in food.

The research showcases the potential for this AI-powered sensor technology to expand into diverse applications.