The Center for Data Innovation spoke to Robert Heinecke, co-founder and chief executive officer of Breeze Technologies, a Hamburg-based company using the Internet of Things (IoT) to monitor air quality. Heinecke talked about the value of sensor data for understanding the environment and how modern data analytics software helps researchers do more with environmental data.
Nick Wallace: Breeze Technologies uses sensors to monitor air quality, both indoors and outdoors. Let’s start with the outdoors—what can outdoor sensors tell us that traditional meteorological methods cannot?
Robert Heinecke: Well if you look at what cities have been doing to monitor air quality, you have those huge containers you see at the roadside that do measure the components of air quality, but they only do that for five to ten places in a city. If you have a metropolis like Hamburg, where we are based, and which has about a dozen measurement stations in a city of 1.7 million people, then this tells you basically nothing about what the air quality really looks like except in the areas where those sensors are deployed. For most citizens in the city, that data does not have any relevance, because the measurement stations are too far away, and the reality citizens in remote parts of the city will face can be quite different from what’s shown in the official data.
Our approach is to have smaller, cheaper sensors deployed throughout the city, and to use this hyper-local data to drive a more objective public discussion about air quality. With scandals like dieselgate, people are unsure about what is around them in the air they breathe, and not much of a way for them to tell because we lack the data. We want to give people a tool to find out about what is happening around them.
We also want to help cities to design their current processes to improve air quality in the city more efficiently. If you only have ten to twelve measurements, and you start a program to improve air quality, you cannot see how well your measures are performing for most of the city, because you only have these few limited spots where you can see the results. Our idea is to use approaches like big data and machine learning that put us in a position where we can say, “this measure that this city tried to improve air quality in this district has performed well at the local level, so we can spend more money on this measure to roll it out in the rest of the city.”
Wallace: Your outdoor sensors are specifically aimed at smart cities. What can a smart city do when air pollution starts to rise in a particular area?
Heinecke: There are multiple things they can do. They can plan building interventions, so when constructing new roads or walkways, or when renovating the existing ones, they could use technologies like photocatalytic asphalt, which contains titan oxides that clean the air using the power of the sun. There are tools like moss walls or algae walls, which you can just put on public facades and then clean the air.
With a data-led approach, the solutions available to cities become more systemic. For example, in London they are using the data to increase or decrease the congestion charge based on current air quality levels. This is a very smart approach, because it does not punish people for driving when there is no problem, but on days when air pollution is very bad, you can use this fee-based approach, based on actual pollution data, to incentivize people not to use their car to get into the city.
Paris, on the other hand, on days when the air quality gets really bad, they make the subway and the buses free to incentivize people to use public transport. This is also a discussion in Germany. You’ve probably heard that Germany faces many infringement proceedings from the European Union, in part because cities often do not comply with European air quality rules. There are some German cities that wanted to try out completely free public transport, but you have taxpayers associations and groups like that who complain about the cost and call for smarter alternatives. So why not follow the example of Paris, deploy sensors throughout the city, and use that data to tell citizens, “ok tomorrow’s forecasts are really bad, so public transport will be free tomorrow,” but not on days where there is no problem at all.
Wallace: What’s the value of putting air sensors inside buildings?
Heinecke: Look at how buildings have changed over the past 30 or 40 years. We started with just building office buildings and putting as many people inside them as possible in the hope they will be as productive as possible. Then we moved to becoming more energy efficient, trying to put in energy saving measures like insulation. Nowadays, we’ve found that we’ve really insulated people in modern buildings—there’s no exchange of air with the world outside, you can’t open the windows, the air in the building is kept inside to keep the warmth and energy in, but this leads to problems like “sick building syndrome,” because our buildings don’t breathe anymore. That really impacts the health and productivity of the people sitting in those buildings.
This is why there is now a new way of thinking about building construction and maintenance, with new building standards like the Well Standard, which forces corporations that want to be certified to care about indoor air quality. How much particulate matter is there in the air? How high are the ozone levels? How much CO2 accumulates before it gets transported outside? To do this efficiently, to care about indoor air quality, you need to give the facility managers and the real estate companies data about how their buildings are performing. You need this data for certification under building standards, but managers and companies also need the data to react to complaints from their coworkers and rentees, and then to find measures that will improve the situation.
Wallace: Do you draw on any other data, like publicly available weather and climate data, when analyzing the sensor data?
Heinecke: Yes, this is exactly what we are doing. We use publicly available climate and weather data from the existing measurement networks in cities. With indoor data, we integrate with building management systems. On the one hand we integrate as much external data as possible to enrich our platform and our data, and become better at calibrating our own sensors, as well as to provide insights for our customers about what is working well to improve the situation. On the other hand, we see ourselves as a vertical expert system on air quality and climate data. If you look at the performance of a building and a city, there is a lot more to consider than just air quality and climate, which is why you need horizontal integration platforms like smart buildings systems and smart city systems, where all these expert systems come together and you can provide another level of insight on top.
Wallace: You mentioned machine learning—what role do you think predictive tools like artificial intelligence in your field in the future?
Heinecke: AI is going to be very important, because there is just so much data to deal with nowadays, and there is going to be more and more as we deploy more and more sensors in the environment. Our buildings and our cities get smarter every day, and there just is not enough manpower in the offices dealing with this data to deal with it efficiently. You need to have tools like artificial intelligence and machine learning to deal with this data effectively, and to find the sweet spots and the gems inside the data where you can generate additional value for citizens and users of a building.
We always try to build scenarios for each customer, which we compare to what has worked for other customers in the past, as well as what we know from scientific studies. Based on these scenarios, we identify the measures that we think are going to perform most efficiently and effectively for our current customer, and we give them ideas like “the top five things to do for your city,” or your neighborhood, or your building.
We also provide insights on how big the current problem is. We use scientific and medical studies, such as a recent one that quantifies the financial impacts and productivity losses of bad air quality, and then we match this with the data we get from our customers. So we say to the building manager, “we expect your co-workers to currently be 25 percent more sick than they need to be, so you have 25 percent more sick days than necessary, and you can implement these measures to bring down the rates of sickness.”