Usual disclaimer that MIT's PR dept is really effective (I've stopped paying attention them because of so much hype), and there's quite a chasm between a proof of concept and a mass-market device.
That one was supercharged by the Media Lab's PR organ. But even without that assistance, the regular MIT "news" office is head and shoulders above its peers in making the trivial appear transformational and the outstanding appear...boring.
My favorite was a publicity piece about a new proof for the behavior of higher-order manifolds with some absurd justification as to why you should care, something like how it would revolutionize battery electrodes or something like that.
edit: looks like the "particle module" is just a LTE/BT communication module, so really there don't seem to be any environmental sensors other than for temperature.
I don’t know if this is the device to do it, but the premise is solid: create awareness that air pollution is a real thing affecting specific areas in which you are breathing right now, not just an abstract bad thing that may or may not be happening somewhere to someone.
I'm debating buying one for myself so I can judge whether air pollution is effecting my asthma. You can find AQI for a lot of cities but it appears the calculation often isn't done with local devices. Satellite imaging?
Naively searching for every lines of the bill of materials gives me around $300, with a handful of items being around $50 and everything else under $5. I'm sure a hobbyist could find the parts for cheaper, especially things like individuals connectors that I found for around $5. I've seen them go as low as $0.25 when bulk purchased.
Disclaimer: it's possible that some SKUs led me to the wrong products. I am not into electronics. I encourage you do to the same exercise before drawing conclusions.
Not to undermine the DIY aspect which could be fun, but it looks like you could buy an Atmotube for about half as much (if you take the research survey) — https://ukstore.atmotube.com/. Though it doesn't register noise levels.
They can also build you an Atmocube which can measure noise and a few other things for $200-$500 based on what I found online.
Impressive, I’ve been wanting to build a portable version of this for a while now. It would look like a smart band but integrate environmental sensors and the users would then be aware of how their environment is affecting them. They would also have the option of contributing their data to a collective dataset.
The key word -- which is missing from the PR title, but is in the paper title -- is "calibration". The novelty is there. Apparently this is a big problem due to varying humidity levels.
I had a similar DIY project, also using some Alphasense sensors, + some cheaper metaloxide ones. It was somewhat promising, but it needs long-lasting colocation studies and, after I moved to Sweden, I lost any support from the local authorities and gave up.
https://bochovj.wordpress.com/tag/air-quality/
there is an old man in Berkeley who has made very good money for more than thirty years, designing and selling these. No, he does not want to "open source" it (I tried).
This sounds highly implausible for anything other than very crude and not-very-useful measurements. Quantifying air pollution is a fairly hard problem, chemically speaking. The composition of particulate matter is highly diverse as it may arise from a wide variety of sources, i.e. agricultural, industrial, wildfires, diesel engines, etc. Just looking at the particulate PM2.5/PM10, broadly speaking there's the organic carbon fraction and the inorganic metal fraction. The former is highly complex, e.g.:
> "The considerably increased chromatographic resolution in GC×GC [gas chromatography] allows separation of many UCM [organic carbon] compounds while the TOFMS [mass spectrometer] supplies mass spectral data of all separated compounds. However, the data sets are getting enormously complex. In a typical PM2.5 sample from Augsburg more than 15,000 peaks can be detected... "
Some particulate matter may have a heavy metal fraction, some may not and that's also not easy to determine (but was a major factor in leaded gasoline pollution). Here's a sample of the kind of work that has to be done to get reliable measurements:
> "...using quadrupole inductively coupled plasma – mass spectrometry (q-ICP-MS). We report improved measurements of key aerosol elements including Al, V, Cr, Fe, Ni, Cu, and Zn in airborne coarse particulate matter (PM10)... This technique was used to determine the elemental composition of over 150 PM10 samples collected from an industrialized region in Houston, TX."
On top of that there's nitrogen oxides and PAN, ozone, etc. The only relatively inexpensive recent innovations seem to be the use of drones to collect samples for lab analysis (would have been useful in East Palestine).
Getting accurate measurements of all the species involved in air pollution requires a modern analytical lab packed with equipment that costs hundreds of thousands of dollars and highly trained technicians to operate. The press release and snippets from the paper don't address such important details at all.
Most definitely it's hard to dig into the actual composition of air pollution. But "how much PM2.5 is in the air", while definitely crude, is still extremely useful. That said, decent sensors for PM concentration are already pretty cheap (<$100 for a module; ~$100 for a full package), so unless the price is going way down it's unclear how useful this particular innovation is.
A lot of these cheap, portable sensors "cheat" in one way or another. E.g CO2 sensors that assume they'll be subject to open air once every few days, using that minimal CO2 value as a baseline. So while they give numbers which can be useful if the user is aware of its limitations, they're far from being analytical tools.
In contrast PM is pretty easy to get right. Shine laser, count particles.
Not only that, but they are deployed on top of automobiles, where the air quality in the surrounding air is going to be worse to begin with. I understand the benefits of using GPS, but I don't see where they mention this being an issue.
We built those and got them running one evening at a Meetup in Erlangen - looks like they had to switch to a temperature/humidity sensor that requires soldering the pins on, because the ones we used back then came ready to plug in, not a soldering iron in sight.
How does this compare to something like an Airgradient, which you can put together for like $50? More detail about what constitutes the PM2.5 or PMn particles you're measuring?
Yes. But I can't see that, the sensor will still output values regardless of being clogged up. Not sure if I can open the device without destroying it, haven't tried. The power runs all the time as long as the USB cable is connected, but it's only a Serial-USB converter.
33 comments
[ 1.0 ms ] story [ 66.5 ms ] threadThe Open Agriculture fraud is a living example: https://gizmodo.com/mit-built-a-theranos-for-plants-18379682...
My favorite was a publicity piece about a new proof for the behavior of higher-order manifolds with some absurd justification as to why you should care, something like how it would revolutionize battery electrodes or something like that.
PS: "living example" was a nice touch.
https://doi.org/10.1016/j.atmosenv.2023.119692
This lists a "particle module", a temperature sensor and some other seemingly irrelevant sensors (GPS, accelerometer).
[1] https://github.com/MIT-Senseable-City-Lab/OSCS/tree/main/Bui...
edit: looks like the "particle module" is just a LTE/BT communication module, so really there don't seem to be any environmental sensors other than for temperature.
https://github.com/MIT-Senseable-City-Lab/OSCS
Link to assembly guide: https://github.com/MIT-Senseable-City-Lab/OSCS/blob/main/Bui...
Link to bill of materials: https://docs.google.com/spreadsheets/d/1-fR-0hTxHKbjaRf8DbH6...
Naively searching for every lines of the bill of materials gives me around $300, with a handful of items being around $50 and everything else under $5. I'm sure a hobbyist could find the parts for cheaper, especially things like individuals connectors that I found for around $5. I've seen them go as low as $0.25 when bulk purchased.
Disclaimer: it's possible that some SKUs led me to the wrong products. I am not into electronics. I encourage you do to the same exercise before drawing conclusions.
They can also build you an Atmocube which can measure noise and a few other things for $200-$500 based on what I found online.
> "The considerably increased chromatographic resolution in GC×GC [gas chromatography] allows separation of many UCM [organic carbon] compounds while the TOFMS [mass spectrometer] supplies mass spectral data of all separated compounds. However, the data sets are getting enormously complex. In a typical PM2.5 sample from Augsburg more than 15,000 peaks can be detected... "
https://www.sciencedirect.com/science/article/abs/pii/S00219...
Some particulate matter may have a heavy metal fraction, some may not and that's also not easy to determine (but was a major factor in leaded gasoline pollution). Here's a sample of the kind of work that has to be done to get reliable measurements:
> "...using quadrupole inductively coupled plasma – mass spectrometry (q-ICP-MS). We report improved measurements of key aerosol elements including Al, V, Cr, Fe, Ni, Cu, and Zn in airborne coarse particulate matter (PM10)... This technique was used to determine the elemental composition of over 150 PM10 samples collected from an industrialized region in Houston, TX."
https://www.sciencedirect.com/science/article/abs/pii/S00032...
On top of that there's nitrogen oxides and PAN, ozone, etc. The only relatively inexpensive recent innovations seem to be the use of drones to collect samples for lab analysis (would have been useful in East Palestine).
Getting accurate measurements of all the species involved in air pollution requires a modern analytical lab packed with equipment that costs hundreds of thousands of dollars and highly trained technicians to operate. The press release and snippets from the paper don't address such important details at all.
In contrast PM is pretty easy to get right. Shine laser, count particles.
Data available at luftdaten.info
https://sensor.community/de/sensors/airrohr/