1.1.3 - Human Activity Recognition Using Distributed Colorimetric Chemical Sensors Based on CO2 and Formaldehyde Inks and Interpretable Machine Learning
- Event
- EUROSENSORS 2026
2026-09-06 - 2026-09-09
Zurich - Band
- Lectures
- Chapter
- Chemical Sensors
- Author(s)
- I. Benito-Altamirano - Universitat Oberta de Catalunya (Spain), Y. Chen, O. Casals, C. Fabrega - University of Barcelona (Spain), M. Gonzalez-Gomez, J. D. Prades - Technische Universität Braunschweig (Germany)
- Pages
- 44 - 45
- DOI
- 10.5162/eurosensors2026/1.1.3
- ISBN
- 978-3-910600-12-6
- Price
- free
Abstract
This work presents a novel approach for human activity recognition based on distributed colorimetric chemical sensors using inks selective to carbon dioxide and formaldehyde. Sensor nodes were deployed in a pilot residential environment to capture spatial and temporal variations in indoor air composition during daily activities. Experiments followed a structured protocol and involved up to sixteen participants. Distinct activity dependent chemical signatures were observed. An interpretable machine learning model was used to classify activities from sensor data. Results demonstrate robust and privacy preserving activity recognition based on environmental chemistry.