SmellControl: AI-driven nano-sensor system for real-time virus detection in seeds

Overview

SmellControl introduces a groundbreaking approach to seed health testing through its AI-driven nano-sensor system. Utilizing volatile organic compounds (VOCs) emitted from infected seeds, this system provides a real-time, non-invasive solution for detecting active viral presence. Unlike traditional methods, which are often time-consuming and destructive, SmellControl leverages machine learning to interpret chemical signatures, offering a faster, scalable, and sustainable diagnostic tool suitable for labs, greenhouses, and field applications.

Technical specifications

The core of SmellControl's technology lies in its multi-channel nano-sensor system, which collects raw gas sensor data and processes it through a machine learning model based on large language model (LLM) techniques. The system features:

• A nanohybrid sensor array capturing 64-channel gas responses with additional temperature and humidity data, creating 66 features in total.
• A cloud-based learning architecture for continuous model improvement through feedback loops and new sample data integration.
• AI models such as K-NN, Gradient Boosting, and CNNs to optimize classification performance.
• Real-time data processing with each 5-hour session generating approximately 6,000 labeled data points.

Technology readiness level

Currently at Technology Readiness Level 4, SmellControl's solution is functional at the prototype stage. The system is undergoing further validation through data collection and controlled lab tests to enhance its VOC libraries and refine its AI models for better classification performance. This ongoing development aims to achieve a scalable, AI-powered screening tool for global seed health.