Researchers have created an AI-based system designed to replicate the human sense of smell to detect and monitor toxic gases in real-time. This innovative technology combines sophisticated artificial neural networks with a network of sensors to quickly locate harmful gases such as nitrogen dioxide, which can lead to serious respiratory health issues.
At the University of Virginia School of Engineering and Applied Science, researchers have developed an AI-driven system that emulates the human sense of smell for the real-time detection and tracking of hazardous gases. By leveraging advanced neural networks alongside a sensor network, the system swiftly identifies dangerous gases like nitrogen dioxide (NO2) that can significantly affect respiratory health.
According to the World Health Organization, outdoor air pollution, including NO2, is responsible for about 4.2 million premature deaths globally each year, primarily due to respiratory illnesses such as asthma and chronic obstructive pulmonary disease (COPD).
This research was recently highlighted in Science Advances.
Graphene-Based Sensors Imitate Human Olfaction
The advanced system utilizes nano-islands of metal catalysts on graphene surfaces. This device acts like an artificial nose, responding to specific toxic gas molecules. When nitrogen dioxide molecules attach to the graphene, the sensor’s conductivity changes, enabling it to detect gas leaks with remarkable sensitivity.
“Nano-islands of metal catalysts are small clusters of metal particles placed on a surface like graphene, enhancing chemical reactions by increasing the area available for gas molecules to interact and allowing the precise identification of toxic gases,” explained Yongmin Baek, a research scientist in the Department of Electrical and Computer Engineering leading the sensor development.
Kyusang Lee, an associate professor in electrical and computer engineering as well as materials science engineering, and a key researcher on the project, stated, “By merging AI with cutting-edge gas sensors, we can accurately detect gas leaks in large or intricate spaces. The artificial olfactory receptors detect minute changes in gas concentrations and relay that information to a nearby computing system, which employs machine learning algorithms to identify the leak’s source.”
Neural Network Enhances Sensor Arrangement
The system’s artificial neural network interprets data from the sensors in real-time, ensuring optimal sensor placement for effective coverage. This optimization is supported by a “trust-region Bayesian optimization algorithm,” a machine learning method that simplifies complex issues into manageable sections to identify the best sensor locations. This approach reduces resource use while providing quicker and more precise gas leak detection.
Ph.D. student in electrical and computer engineering, Byungjoon Bae, remarked, “Our AI-based system can enhance safety in industrial locations, urban areas, and residential neighborhoods by continuously monitoring air quality. It’s a significant advancement in reducing long-term health risks and safeguarding the environment.”
The paper, titled “Network of Artificial Olfactory Receptors for Spatiotemporal Monitoring of Toxic Gas,” was published in Science Advances. The research team comprises Yongmin Baek, Byungjoon Bae, Jeongyong Yang, Wonjun Cho, Inbo Sim, Geonwook Yoo, Seokhyun Chung, Junseok Heo, and Kyusang Lee, collaborating across institutions including the University of Virginia and Ajou University.
This research received funding from the Industrial Strategic Technology Development Program (20014247 and 20026440) by the Ministry of Trade, Industry, and Energy (MOTIE, Korea), the National Research Foundation of Korea (NRF), and the US Air Force Office of Scientific Research Young Investigator Program (FA9550-23-1-0159), with additional support from the National Science Foundation (NSF ECCS-1942868 and NSF ECCS-2332060).
