Awareness for the invisible: Predict healthier journeys from air pollution data

Posted on by Phil Wornath

Premise

Creating healthier cycling commutes using crowdsourced IoT

GitHub: https://github.com/philffm/FreshAir

Figma Concept: https://www.figma.com/proto/

Synopsis

The global climate and health crisis is affecting everyone and from 2030 and 2050 is expected to account for “250 000 additional deaths per year.” (World Health Organization (2021)

Air pollution, which poses a major challenge in most emerging economies, is an invisible, yet integral part of this crisis

But how can we make this invisible health threat more visible for society – and in the next step make them learn & reflect on it?

My proposal consists of 3 iterations that have gradually emerged from my observations, evaluations and reflections, as well as a recommendation on how the project could be proceeded in the graduation project.

Substantiation

My overall topic deals with the question on how we might raise awareness in society on the invisible air pollution – and  in the next step let people on bicycles choose the least polluted commute using crowdsourcing-enabled IoT technology.

Within the course “State of the Art Technology” I was focussing on the first part of this question, on how we might create awareness using an IoT device, as well as building the technical foundation for what comes next: The graduation project.

Microcontrollers for IoT projects have been getting more affordable and accessible to consumers than ever before. During this course, I took the opportunity of getting prepared for the technical challenges and obstacles of working on such physical prototypes. I did this by exploring the technology, revised the coding syntax of C++ (Arduino), finding out how suitable and reliable today’s consumer sensors are for measuring such data and developed an MVP technology stack for as means of foundation for my graduation project.

Testing with potential users sparked new ideas on how to increase privacy and how to create more meaningful interactions in the final prototype. Collecting air pollution data in the field and evaluating it created demand for testing more sensors for potentially increased precision in the graduation project.

Overview of prototypes

Iteration 1: Paper Prototypes & TinkerCad

This first iteration consisted primarily of building a rough story line based on the given “How might we…”-question  as well as prototyping the required sensors for a technical foundation.

Paper prototype 1: A rough concept

Very few products experienced market maturity in the first iteration. And this is especially true for IoT prototypes: Wildly plugging cables together in the first iteration doesn’t do much good.

That’s why I started from scratch – on paper illustrating very roughly how to bring the data to life in public, while addressing multiple potential users at once.

Technical paper prototype 1: Playing with TinkerCad

Since my hardware was still stuck in customs, I decided to start on a sheet of paper instead of starting immediately with the hardware itself.
Using that approach, I was able to map out a concept, draw the electrical circuits and simulate what I wanted to implement later, thereby work more goal-directed in the further iteration phase.
TinkerCad allows early-stage protyping of IoT prototypes – and supports the ideation process with components you may not have thought of yet
With TinkerCad I discovered a new tool that allows me to constructs schematics for my physical IoT prototypes – basically the Figma for electrical engineers (or least for those who want to become electrical engineers, since real pros seem to make use of tools like KiCad)

Iteration 2 – The first test run: Reading data from sensors locally

After the hardware arrived, we were ready to go.
Since I used the first iteration to visualise the connections beforehand, it was easier for me to implement the second, physical iteration.
The hardware setup consisted of the following components:
  • ESP8266 Module (Works like an Arduino Uno)
  • WiFi
  • CCS811 CO2 Sensor
  • Serial logging via. USB
Iteration 2: Functional test. After finding a defective cable as the main cause for the sensor not reporting any data, we were all set.
The connection of the sensors succeeded after two attempts and the integrated Arduino debugger allowed insights into the inside of the hardware.
Does it work? The Serial Debugger is a great tool to make your electronics talk to you.

Feedback gathered from internal testing was primarily about the aspect that that there should be a direct feedback given from the IoT device – because: The only way to get data from the device so far was by reading its serial output.

But what makes the most sense here? Outputting the live data in numbers? Lights? Using voice output?

Iteration 3: WiFi Connection,  Google Sheets & LED

A key challenge was that the data could currently only be read out locally / offline.
From my experiences with Python I knew that there is a library for pretty much everything: And since I was familiar with Google Sheets API – and found the idea exciting to watch a document being filled by ghost, I decided to make this my first mean of collecting data. Since there was no direct library, every 5 seconds the ESP8266 module sends an HTTPS request with the current  air quality data.

Logging the data – from the micro controller straight into Google SheetsIn addition, I added an LED that glows green to red on the spectrum from 400 to 800 PPM (CO2).

The first user interface is already in place: The LED is aimed to inform cyclists about bad air conditions while riding.
User Testing in the classroom – Breathing into the sensor demonstrates the basic functionality of the device

During the user testing we found that the LED is already a very suitable choice to visualise the air quality in context. What was often mentioned is that lights were “much less distracting than, for example, a display with numbers”. This is an important factor to consider when it comes to a gadget that is used while riding, where driving safety is a top priority.

However, one important point that was mentioned was about learning from the data: Users would like to see the data available also after the ride – and possibly make them available also to others.

Testing in the wild

While I have already been tracking data with the sensor in my backpack due to Dutch weather conditions 🌧  (and a device that was not yet rain-proof), I also wanted to start a first test ride  with the device mounted on the bike.

First test in the wild: The sensor seems to detect a higher CO2 concentration with cars passing by

After testing the (now exposed) device for 10 minutes outside, one of the first cyclist (and potential user) at a traffic light approached me:

“What’s that device you have there?”

I replied “What would you guess?”

He initially suspected a distance meter for overtaking cars. A good thought – my hint that it is about visualising something invisible finally brought him to air pollution.

This first spark of interest from someone external raised my curiosity in conducting more user testing with people on the street (and I can’t stress this enough, since forcefully asking fellow students, friends and flat mates for feedback is something I’ve grown way to tired of after several years of “you have to conduct user testing” – especially during corona, where research was rather experimental and the ability of approaching random people in public needs to be relearned by many.).

Therefore I’m looking forward to testing the next prototype iterations in real life, making the device as eye-catching as possible, and also incorporating a way for user feedback that doesn’t require my presence.

Exciting to see: The sensors do indeed output higher values for cars passing by, even with this 3$ sensor.

Outliers: The values skyrocket as soon as I enter my apartment. This is probably because cheaper sensors work with air pressure – unlike higher-quality sensors such as the PMS5003, which works with laser particle measurement and is due to be tested in the next stage.

Future Concept – What is this all about?

What I tested here as part of SAT is just one piece of the puzzle of an entire ecosystem. Want to see more? Here are slides.

View presentation on Figma

Conclusion

A major issue that has also come up in feedback with potential users, is the storing of data, as well as privacy.
For the sake of simplicity, I used Google Sheets to capture the data for this project at the moment.
The users surveyed would prefer to have their data anonymized – or pseudonymized – and, if pseudonymized, to be able to specify an area (such as a radius around their home address) whose location data is not recorded.
Since no location data has been collected yet, I think using Google Sheets is viable for now. The aim however is to find a more secure way of storing data during the graduation project that would work in a  decentralised and pseudonymised manner. For instance Projects such as IPFS (Interplanetary File System) in conjunction with IoT devices sound promising, though not yet widely deployed and more research needs to be done.

Another valuable insight I will further investigate and focus on during my graduation project is the user interaction.

So far, the focus has been primarily on one user – and one context: the cyclist himself – while riding. But what about interaction with other people? Who else could be part of the ecosystem by (first-time) interacting with it, and explore or contribute data to it in what ways?

I’m excited about the experiences and insights I’ve gained – and I’m looking forward to learning more in the graduation project.

Sources / Inspiration

Beedham, M. (2020). Every navigation app should have Cowboy’s air-quality ebike route feature. TNW. https://thenextweb.com/news/every-navigation-app-should-have-cowboys-air-quality-ebike-route-feature
Benet, J. (n.d.). IPFS-Content Addressed, Versioned, P2P File System (DRAFT 3).
Gatto, N. M., Henderson, V. W., Hodis, H. N., st. John, J. A., Lurmann, F., Chen, J. C., & Mack, W. J. (2014). Components of air pollution and cognitive function in middle-aged and older adults in Los Angeles. NeuroToxicology, 40, 1–7. https://doi.org/10.1016/j.neuro.2013.09.004
Reid, C. (2020). Bicycling Booms During Lockdown—But There’s A Warning From History. https://www.forbes.com/sites/carltonreid/2020/05/01/bicycling-booms-during-lockdown-but-theres-a-warning-from-history/
Johnston, S. J., Basford, P. J., Bulot, F. M. J., Apetroaie-Cristea, M., Foster, G. L., Loxham, M., & Cox, S. J. (2016). IoT deployment for city scale air quality monitoring with Low-Power Wide Area Networks.
Quick, M., & Posavec, S. (2015). Air Transformed. https://miriamquick.com/air-transformed
Wesseling, J., Hendricx, W., de Ruiter, H., van Ratingen, S., Drukker, D., Huitema, M., Schouwenaar, C., Janssen, G., van Aken, S., Smeenk, J. W., Hof, A., & Tielemans, E. (2021). Assessment of pm2.5 exposure during cycle trips in the netherlands using low-cost sensors. International Journal of Environmental Research and Public Health, 18(11). https://doi.org/10.3390/ijerph18116007
World Health Organization (WHO). (2021, October). Climate change and health. https://www.who.int/news-room/fact-sheets/detail/climate-change-and-health
WNYC. (2017). Biking and Breathing. https://www.wnyc.org/series/biking-and-breathing

Leave a Reply

Your email address will not be published. Required fields are marked *