Using artificial intelligence to improve air quality

The challenge: Improving on state-of-the-art measurement and evaluation

“We have a clear mission and that’s to use our sensors to help provide transparency on air quality and sustainably improve it,” explains Robert Heinecke, founder and CEO of Breeze Technologies. The company’s sensors are many times smaller than conventional measuring stations for air pollutants, which are the size of SUVs and located on busy roads. However, it’s not their size but the technology within that makes the difference here. Their sensors measure the level of harmful substances in the air, such as carbon dioxide, nitrogen dioxide, or particulate matter, just as reliably and accurately. Then they automatically upload the data to Microsoft Azure for evaluation. Founded in 2015, the company is still relatively young—but the hundreds of sensor locations and more than 10,000 measuring points it has around the world point to rapid growth. And due to increasing demand for air quality measurements, Breeze’s customer base is growing, too. “Our customers aren’t all municipal authorities. Lots of companies engage our services, too, for example to test the air quality where their offices are located. Companies in, say, the chemical industry are very keen to know what the air quality is like in their facilities,” Heinecke says. “We also make the data we collect available on our Citizen portal, where we present it transparently. Our data helps to substantiate and objectify the public debate.” 

But the services that Breeze Technologies offer extend beyond measuring air quality. The company also advises its customers on how they can make lasting improvements to air quality. “Every customer can access their own data and analytics for themselves. The next step is for them to determine how to translate that information into specific action. We advise our customers and help them choose the right air pollution control measures from over 3,500 different options,” Heinecke says. 

But for Heinecke and his team, the revolution in sensor size and automated data evaluation using the Azure Cloud are just one piece of the puzzle. “As a start-up, we’re always looking to innovate. The quality of data that our sensors can achieve on their own is limited. Using artificial intelligence is the only way for us to further enhance the accuracy of our measurements, which is why we pool them all centrally in the Azure Cloud. Known as cloud calibration, this delivers even higher quality data,” he explains.

The solution: AI eliminates disruptive factors and provides insight into the future

Breeze Technologies has transferred the previously used measurement methods in all their complexity from the measuring stations to the Azure Cloud—now, quality assurance, plausibility checks, and calibration all take place on Azure. In its own AI experimental laboratory, the company develops solutions that it can use to obtain even more precise measurement data and to detect potential problems at an early stage. “Our sensor units are sensitive to certain environmental influences, such as temperature. AI in the cloud filters those parameters out of the data, which makes the final measurements more accurate,” Heinecke says. Calibration technology among other things based on Azure Functions converts the data collected by Breeze into other scientific variables. The final measurement is generated on the basis of various statistical methods and with the help of artificial intelligence.

But Breeze Technologies has even bigger ambitions still. “We are well on the way to equipping our sensors with predictive maintenance technology. In the future, they’ll be able to send out an automatic notification once they’re likely to need replacing soon.” That will mean no more measurement errors or failures of entire sensors caused by defective components.

“Azure offers an enormous range of possibilities. In the future, for example, we’ll be able to incorporate weather data and associated traffic volumes. That would let us predict whether many people will be driving into the city on a given day, which would push up traffic volumes and make the air quality worse,” Heinecke explains. In turn, municipal authorities could offer discounted tickets for public transport on those days or take other air pollution control measures to directly counteract the effects. 

Breeze Technologies wants to make its air quality data available to other industries—such as insurance companies, real estate companies, health care providers, or health insurance companies—that could all stand to benefit from it. Microsoft is now supporting the company as part of its AI for Earth initiative, through which Microsoft makes the Azure Cloud Platform and its AI tools available to project teams working to solve global environmental challenges.