AI Smarts Let This Face Mask Adapt Its Filtration to Exercise, Pollution Levels, and More

Opening and closing its pores based on nearby pollution and the wearer's respiration, this smart mask balances comfort and protection.

An international team of researchers have developed a filtering face mask, which adjusts exactly how much filtration it offers according to everything from surrounding air pollution to how much exercise you're doing — thanks to a machine learning algorithm.

"The recent emergence of highly contagious respiratory disease and the underlying issues of worldwide air pollution jointly heighten the importance of the personal respirator," the researchers explain of the inspiration behind the project.

"However, the incongruence between the dynamic environment and non-adaptive respirators imposes physiological and psychological adverse effects, which hinder the public dissemination of respirators. To address this issue, we introduce adaptive respiratory protection based on a dynamic air filter (DAF) driven by machine learning (ML) algorithms."

This face mask learns to balance filtration efficacy with airflow needs, thanks to a link to a machine learning system. (📹: Shin et al)

The DAF looks at first glance like any other face mask respirator — except for two tubes which run down from a pair of rings at either side. As the filtration required changes, these tubes open and close the rings — stretching or relaxing a dynamic filtration fabric to increase or decrease airflow accordingly. If you enter a polluted environment, air flow is sacrificed for filtration efficacy; if you start exercising and breathing heavily, air flow is increased at the cost of filtration.

The mask links to a small wearable system with an air pump, sensor, and microcontroller; in its present incarnation, however, the machine learning part of the system relies on a more powerful external computer to which the wearable portion communicates over a wireless link.

"The stretchable elastomer fiber membrane of the DAF affords immediate adjustment of filtration characteristics through active rescaling of the micropores by simple pneumatic control, enabling seamless and constructive transition of filtration characteristics," the researchers explain.

"The resultant DAF-respirator (DAF-R), made possible by ML algorithms, successfully demonstrates real-time predictive adapting maneuvers, enabling personalizable and continuously optimized respiratory protection under changing circumstances."

The next step: Building a next-generation DAF-R which doesn't rely on pneumatics to stretch and relax the filtration fabric, thus reducing the size and complexity of the mask.

The team's work has been published in the journal ACS Nano under closed-access terms.

Main article image adapted from ACS Nano 2021, DOI: 10.1021/acsnano.1c06204.

Gareth Halfacree
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire:
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles