Air quality sensor embeds AI capabilities -

Air quality sensor embeds AI capabilities


The Covid-19 pandemic has stressed the importance of improving indoor air quality. Whether at home, in the office or in public transportation, it is more important than ever to ensure the air we breathe is clean and safe. Bosch Sensortec has developed an environmental MEMS sensor and an AI-based software to reduce the risk of virus infection.

Bosch Sensortec’s BME688

MEMS sensors have made their way into our everyday lives. Nonetheless, to continue to provide value to the customers, Bosch Sensortec is convinced that a stronger focus on hardware-software co-design is required by using advanced algorithms and embedded AI.

For Bosch Sensortec, sensor software will become increasingly intelligent, turning MEMS sensors into more accurate, more secure, and personalized systems that can help the user adapt to any situation. Software adds value not only to the sensor but also to the entire system, said Ralf Schellin, vice president and head of product area MEMS at Bosch Sensortec, at the recent MWS 2021 MEMS Titans Webinar.


With every exhaled breath, small droplets are emitted into the air. If someone is infected, respiratory droplets and airborne transmission can carry the virus to the others. The lower the ambient humidity or temperature is, the longer the aerosols can stay in the air, said Schellin.

The Bosch Sensortec’s BME688 is an air quality MEMS sensor that combines four sensing features – gas, humidity, temperature, and barometric pressure – with AI capability.

The gas sensor can detect Volatile Organic Compounds (VOCs), Volatile Sulfur Compounds (VSC) and other gases such as carbon monoxide and hydrogen in the part per billion (pbb) range.

Although the BME688 sensor does not detect the Covid-19 virus, “it can help significantly reduce the risk of infection by measuring VOCs, humidity, temperature,” said Schellin. By a concentration-controlled ventilation, both the infection risk and the impact can be significantly reduced. “In the end, it gives you a kind of signal to open the window or switch on the air cleaner, or you can set up an automatic system in the area.”

AI-based software

The BME688, or what Bosch Sensortec calls ‘a digital nose’, can detect gases by measuring their unique electronic fingerprint and distinguish different gas compositions. First, the sensor has to be taught about these different gases, which means it needs to be trained using machine learning methods. That’s where Bosch Sensortec’s BME AI-Studio software comes in. It lets customers explore and verify their specific use cases. They can collect gas measurements, train machine learning algorithms and export a final algorithm to be used with the BME library in their project.

“You can investigate your own use cases and train the system for exactly this use case,” said Schellin.

By temperature cycled operation of the BME688‘s gas MEMS, different gas mixtures can be recognized and classified. (Image source: Bosch Sensortec)

In standard configurations, the company said the presence of VSCs is detected as an indicator for e.g. bacteria growth. The gas scanner can be customized with respect to sensitivity, selectivity, data rate, and power consumption. The BME688 has a gas scanner function, and the BME AI-Studio tool enables customers to train the BME688 gas scanner on their specific application, like in home appliances, IoT products or smart homes, the company said.

The software itself is running on the application processor, which is not integrated, said Schellin. “We have the MEMS, we have the ASIC with a certain functionality. But the overall emulation is running on the application processor. It’s just a question of time when it can run on the device itself. We have shown it already with the BHI motion sensor where it’s running on the integrated device. So, it’s not that we are not able to do it. It’s currently not running on it.”

Bosch Sensotec said it is providing an Adafruit-compatible development kit.

3.0 x 3.0

The BME688 sensor is housed in a compact package, measuring just 3.0 x 3.0 x 0.9 mm. Taking the example of the company’s BMM150 magnetometer sized 1.56 x 1.56 x 0.6 mm, Schellin said he is confident the size could be further reduced. Once the physical limit is reached, however, “you add more and more smartness and more functionality, but we’re definitely not there yet.”

From the sensor design and packaging to the embedded AI algorithms, Bosch Sensortec said everything is done in-house to maximize cointegration synergies. If, for instance, “you need to go towards extremely low power consumption, but you don’t know the hardware, you will not read the power level you can reach,” Shellin illustrated. “That does not mean we are not working with partners, but in this case we try to develop everything ourselves.”

>> This article was originally published on our sister site, EE Times Europe.

Related Contents:

For more Embedded, subscribe to Embedded’s weekly email newsletter.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.