‘Seeing’ the air that you breathe
- Shaun Smillie
A new AI-driven air quality monitoring system gives people the power to understand the risks to their health posed by air pollution.
Johannesburg’s air quality has never really been measured systematically. Like many other cities across the globe, scientists have battled to develop cost-effective monitoring systems that provide accurate real time data on air pollution.
This is all about to change, thanks to some home-grown tech and the power of Artificial Intelligence (AI).
For the last couple of years, Wits University, with collaborators from around the world, have been developing and testing a new AI-powered air quality monitoring system, called AI_r.
The study’s pilot saw sensors placed along the M1 highway, close to the University, followed by the next phase of the roll-out and the deployment of over 500 air quality monitoring devices in strategic locations across Gauteng.
AI reducing costs
Traditionally, air quality monitoring has been hampered by the need to use very specialised teams of experts to interpret the data. “This makes it extremely expensive and not even the global north can afford it. So we use AI to do the job of interpreting data,” says project leader, Professor Bruce Mellado from the Wits Institute for Collider Particle Physics. “AI allows us to interpret the data, to make forecasts automatically in real time and to make models based on data – that is where most of the costs lie. We are creating a lot of data and unique data sets that fit into what will become the most sophisticated air quality model based on artificial intelligence, probably in the world.”
Mellado’s own experience as a particle physicist at the Large Hadron Collider at CERN and the training of his students at the international facility, have provided his team with invaluable expertise in working with big data, which is put into use with the air quality monitoring system. He used a similar approach to set up Gauteng’s 足球竞彩app排名 Dashboard, which allowed health officials to forecast new outbreaks during the pandemic.
The AI_r system has attracted international attention and Mellado and his team won the 2025 ODESS Prize in France, beating 350 entries from around the world. The award was made in recognition of the design and deployment of the AI-powered environmental monitoring system.
How to ‘see’ air
In each of the system’s sensors, a laser is used to quantify the concentration of particles found in the air. The data is uploaded through Wi-Fi and the Internet. The system is modular, so the sensors can be upgraded to monitor different chemicals and pollutants within the environment.
AI_r is already providing a snapshot of the air that Joburgers breathe in on a daily basis.
One of the issues that has already been picked up is that Johannesburg has a problem with illegal dump fires, including one that happens regularly in Kya Sands to the north west of Johannesburg. According to Dr Mpho Mathebula from the Wits Department of Psychology who has been studying the effect that these fires have on residents, there have been reports of people experiencing dizziness and battling to breathe, with some even hospitalised.
While work continues to determine what pollutants are released in these fires, the AI_r network of sensors has been able to pick up and track how smoke from the Kya Sands dump fire spreads across the City, covering areas where hundreds of thousands of residents live.
Air pollution kills millions of people every year. The World Health Organization estimates that about seven million people die prematurely from illnesses attributed to household air pollution caused by fuels and kerosene used in cooking.
“We are not trying to gather new evidence to prove how poor air quality impacts on the health of populations. That is already known. What we are trying to do is to find tools that can get the government to understand how bad the problem is and which areas are most affected,” says Wits School of Public Health Professor Mary Kawonga, who is also part of the project
In the future, the amount of real time data generated by the monitoring system could allow forecasting. An app on a phone could, for instance, warn of a coming spike in poor air quality in a particular suburb hours before it arrives.
“The forecasting is done not just for time or as a function of time but it is also done also as a function of space,” says Mellado.
While such sensors can help provide data that helps identify problems, for Mathebula, such a tool can ultimately help citizens in their fight for a healthier future. “This tool will empower the community to advocate for cleaner and healthier environments,” she concludes.
- Shaun Smillie is a freelance writer.
- This article first appeared in?CURIOS.TY,?a research magazine produced by?Wits Communications?and the?Research Office.
- Read more in the 20thissue, themed #Thrive, which explores what it truly means to flourish — across a lifespan, within communities, and on and with our planet.