In Africa, air quality has emerged as a critical public health concern. More people die from poor air quality each year in the world than from HIV, TB, and malaria combined. And that’s only the very beginning. People who are exposed to air pollution have headaches and fatigue, which reduces their productivity.
India is one country where the air quality is terrible. India’s gross domestic product is impacted by bad air quality to the tune of almost US$100 billion annually.
It is well established that bad air quality poses health hazards. However, setting up monitoring stations to take regular measurements has always been expensive.
In addition to being the head of the iThemba Laboratories for Accelerator-Based Sciences, a division of South Africa’s National Research Foundation, and the Institute for Collider Particle Physics at the University of the Witwatersrand, I am a particle physicist. We established the South African Consortium of Air Quality Monitoring, a diverse group of scientists interested in enhancing South Africa’s air quality and having solid relationships with policymakers, as part of our technology transfer initiatives.
We decided to develop the first-ever, reasonably priced, sensor-based, Internet of Things, and artificial intelligence-based air quality monitoring system in South Africa. This system is called Ai_r.
Over 20 years of expertise as particle physicists working with sensors, communications, and artificial intelligence are among the 25 members of our team. Particle physics involves the creation of complex systems involving various disciplines and technologies. Artificial intelligence, data science, electrical and information engineering, and other fields are included in particle physics.
In South Africa, there are just 130 large air quality monitoring stations. Only the area around the station’s air quality is measured. For this reason, to assess air quality over a much larger region, we need dense networks that are affordable and composed of AI devices that are installed all around these stations.
Tens of thousands of these devices will be distributed throughout South Africa, as per our plan.
Using artificial intelligence
Ai_r is composed of several tiny boxes that were manufactured for roughly USD 100. Any building’s window sill may accommodate the boxes, which are used to collect air samples and transmit real-time data to a cloud. Cloud-based artificial intelligence is used for forecasting and modeling.
Machine learning is not a magical art. Scientists use a collection of mathematical instruments to carry out certain tasks. It combines data sets, applies machine learning to the data, and generates autonomous models. That results in massive resource savings.
Artificial intelligence will make predictions in air quality monitoring by analyzing the copious amounts of data generated by these sensors. For instance, Ai_r will be able to inform us about how weather variations affect air quality, allowing us to predict which regions may experience higher levels of air pollution.
It would be extremely difficult to develop a system that could monitor air quality and provide us with predictions at a reasonable cost without artificial intelligence.
Measuring air quality in Johannesburg
Finding the hotspots—areas with the poorest air quality—rather than just the city’s average air quality is the aim of air quality measurements.
A hotspot with extremely contaminated air can be located quite close to a cold spot with cleaner air. We must be able to predict the areas where the air quality will be poor on any given day for public health concerns. We won’t be able to persuade the authorities to implement measures to reduce air pollution in severely afflicted areas until we do this.
About 20 of the new devices have already been distributed in Johannesburg’s Soweto and Braamfontein, and another 120 will be available in the coming months. Tens of thousands of cars pass through both areas of the city each day, increasing the risk of air pollution.