How AI & IoT provides environmental data to combat air pollution

What is the story behind the vision and mission strategy of Ambee ?

Ambee is an environmental intelligence startup that measures, processes and analyses hyperlocal air quality data in real time. Using AI and IoT, Ambee provides location specific, real time environmental data and actionable insights to businesses and consumers to mitigate the effects of various threats from environmental factors, for example air pollution.

The idea for Ambee came about when one of the co-founders Madhusudhan Anand’s six-month old child had life threatening breathing issues due to air pollution. In his quest to understand the problem, Madhusudhan started monitoring air quality in and around his residential community. He found that the nearest air quality monitoring station was located 13 kms away from his house and it suggested the air quality was fine. To get a better sense of air quality in his immediate surroundings, he put together a rudimentary air quality sensor that could give him real time data about the quality of air around his residence. He found that the levels of particulate matter (PM2.5) near his house were touching dangerous levels of 800 ug/m3. This was the genesis of Ambee.

The air quality index in most Indian cities is above the government’s indicated healthy range. In spite of alarming statistics and facts, there is little being done to combat pollution and solve this crisis. There is hardly any reliable data for a common man to understand and gauge the quality of air in his/her immediate surroundings. Ambee was started in 2017 with a mission to measure and democratise access to environmental data and empower the battle against climate change, for future generations.

How are you using data technology for a cleaner environment ?
Till date, various methods have been used to monitor air quality- from large monitoring stations costing hundreds of thousands of dollars, to satellite imagery. So far, none have succeeded to an acceptable level of accuracy and granularity. This is where our technology steps in.

We combine the data from, on-ground sensors, American and European satellites, along with models trained on historical data. A continuous error correction engine is in place to compare data against reference grade stations including eBAMs (Beta attenuation monitors) that not only ensure benchmarking but also accuracy. We also account for specific factors like dust due to construction, traffic, and population density. All of these correlate with air pollution, and the environment. This results in the highest possible accuracy of data, at a high spatio-temporal resolution.

We build data products across a variety of sizes and requirements. We have our API product, which works on a standard subscription model and is used across industries, from large corporations to solo developers. We also build custom products across multiple sectors. In insurance, we’re partnering with large insurers to bring out a product that uses our environmental data to understand risk, which is a first of its kind. We also work with other industries that have environment and climate as a primary focus, from healthcare and pharma, to manufacturing, to aerospace.

How can AI and IoT measure real-time air quality in cities across India ?
As per the Greenpeace report, 22 out of 30 of the world’s most polluted cities are in India.The reality is that there are not enough stations that measure the air quality in India. A base data set is needed for everything from city planning and traffic management to new drug development and clinical trials. There is a big need to solve this problem of lack of data and make this available for everyone.

We solve this problem in three ways :
1. Measure & monitor – With Ambee air sensors and multiple satellite stations we can measure air quality for the entire country round the clock
2. Compute & analyse – Computing to understand patterns in the behavior of pollutants versus weather changes, changes in regulations, economics etc., correlations from continuous emissions and analyzed for understanding sources, places, and more
3. Action & intervention – Alerts during severe environmental distress, actions of small things for social innovation like not using diesel generators to promoting public transport to even turning on your air purifier, telling users when to go for a walk or a jog, take their child for a walk or even helping authorities reduce garbage burning practices across cities to ensure reduction in air pollution.

Ambee’s technology:
Proprietary Algorithms: Spatial Interpolation vs ANN/RNN interpolation over a custom Algorithm:
Spatial interpolation is a way to estimate values of the variable under study at unsampled locations, using point observations within the same region. Statistical methods like Nearest Neighbour, Triangulated Irregular Network, Natural Neighbour, Kriging, Inverse distance weighting, Radial basis functions interpolations, Linear RBFs, Gaussian RBFs are commonly used for solving such problems. However, these methods do not factor in region specific parameters such as weather (temperature), wind speed, wind direction, population density, human activities like traffic, stubble burning -etc. We have developed an in-house training model with years of data to utilise these parameters. This model can predict the air quality at a much higher accuracy helping people get a real time understanding of the air around them. This gives them an opportunity to take necessary actions, create interventions and save time and money.

We have also launched India’s first hyperlocal pollen API that we are seeing a lot of interest in from wellness and therapeutics companies.

Whom are you partnering with in the country – in government as well as the private sector?
We’re working across industries, with large Fortune 500 companies like Airbus and Bosch, as well as with startups. We are also exploring partnerships with governments in different parts of the world.

How are you looking to advocate the cause among the populace and create greater awareness ?
We have seen that awareness comes from information. We have launched a free mobile app that gives near real time hyperlocal air quality, pollen and microclimate data. In addition to this, we have launched two websites www.indianairpollution.com and www.globalairpollution.com that can be easily accessed by people to get a quick overview of air quality in their immediate vicinity. Through our interactions with various stakeholders, we have noticed that when people are equipped with the right information, they are more than willing to make change. For example, very recently, during our interaction with an RWA (Resident Welfare Association), we highlighted the adverse health and environmental impact of diesel generators. The RWA swiftly took action and moved to solar power for backup. On the B2B side, we are working with companies to build sustainable interventions which encourage changes like carbon capture and storage or carbon neutrality, or business models for EV adoption.

Your strategic plans for the future – new tech initiatives and geographic expansion ?
We already work with a combination of large and small enterprises across the world and we expect that to ramp up in the next 24 months. We’re building interesting products across the gamut of environmental intelligence, working with some of our partners.

Any other significant factor you want to highlight ?
I think it’s a good sign that we’re seeing such interest in climate and the environment. There is a lot of brainpower and money now looking to change the way we perceive environmental threats and other factors. For example, we’re accelerated by Techstars (who invested in Uber very early on), Google, and Airbus. This is similar to the waves of tech and funding that entered retail, commerce, and finance. All of this brings with it a deeper understanding of the basic problem, and short and long term solutions to each facet. This time, it’s a problem that could be the end of us if we don’t solve for it aggressively.