The Tech Behind the Mind-Reading Pangolin Dress | Hacker News Reader

The Tech Behind the Mind-Reading Pangolin Dress

spectrum.ieee.org

72 pointsrbanffy18 comments

> Measuring 1.6 millimeters on a side, the single-channel chip integrates an amplifier, an analog-to-digital converter (ADC), and a digital signal processor—and it draws less than 5 microwatts of power.

> With 1,024 separate channels, you can get single finger resolution.

Wow. Anyone who's worked with EEG knows how maddening signal quality issues can be. This puts the amp right at the source, with insane density.

Only downside is you have to shave your head, Professor Xavier style.

I can painfully confirm that data quality issue to be true, any sensor outside the body trying to measure the inside there's always a challenge.

> Only downside is you have to shave your head, Professor Xavier style.

Yes, but you gain the ability to find other mutants!

As an outsider with a bit of EE background, this seems pretty dang impressive. Seems like a potential huge boon for the disabled. Something like this might have made it easier for Dr. Stephen Hawking to communicate.

More info : https://iic.jku.at/analog/pangolin/

Technical yet accessible presentation : https://youtu.be/Gnevq4WFuZU?t=14760

Demo of the moving dress : https://youtu.be/Gnevq4WFuZU?t=18932

TLDW : ( Custom old tech (180nm) integrated chip gets energy from RF harvesting ~5uW -> x100 differential amplifier -> 16 channel - 16 bits ADC (0.5Hz-256Hz) -> ultra low power custom RF transmitter) * 64 pcbs = 1024 channels -> Base station to collect and process the data.

Very cool tech. No battery needed (aka inherently safe). Need to shave your head. Read-only (as opposed to something like neuralink which is read-write).

No battery doesn't always necessarily imply safety. In this case it is definitely safe due to the low power but you probably don't want to walk around a TV transmission ground station holding a giant high gain antenna in your bare hands (this particular scenario is unlikely to happen in real life but it does make for an interesting thought experiment)

This thing serves no purpose if the movement and lights are seemingly random even if they are based on "brain waves". There's no cohesiveness to what it's doing based on the "data". It's just tech for tech's sake, and the result isn't quite as stunning as the creators want us to think it is. It's a nice build, but that's practically where the interest stops for me. Also, it's not a "dress" but more of an exoskeleton. Sorry I can't really spin this any more positively than I did, but it's my honest opinion.

Yeah based on their description of what this is supposed to be doing, we should be seeing some pretty interesting control over the lights and motors. But they mostly just seem to flutter randomly. If this was mounted on a statue instead of a living person would it behave differently? It’s disappointing after seeing so much video I’m not certain the answer is yes.

> This thing serves no purpose if the movement and lights are seemingly random...

It looks neat. It makes for good discussion. It inspires others to think about various ideas. Art serves a purpose, but we all appreciate it differently.

> it's not a "dress" but more of an exoskeleton.

Who made you the arbiter of "dress"? :D

>Who made you the arbiter of "dress"? :D

Who made you the arbiter of when someone can express an opinion??

My opinion is that this is not a dress, it is closer to an exoskeleton.

So, every bunch of circuits and wires I might hook up to some motors and lights is now considered "art"? That's good to know.

This is mostly correct. However, one could add an ML classifier to find patterns in EEG that correspond to e.g. happy vs serious mood. Mood recognition can be done from a camera pointed at the face. If there's any rhyme or reason to the set of 1024 EEG series, the network would learn it.

Just tried submitting this with the title "Tech in Mind-Reading Dress Could Allow Wireless, Batteryless Exoskeleton Control" which does fit, in case you want to consider changing it to that

The designer, Anouk Wipprecht, has some cool other devices as well. One monitors P300 waves (and is styled as a unicorn horn) to help kids with ADHD. https://www.anoukwipprecht.nl/gallery/

Only 5 microwatts of power, and it's still fabbed on 180nm. I'm assuming these "tricks" they speak of wouldn't be ruined if they went to, say, 5nm and reduced the power requirements even further. I hope I'm right because if so that's a true breakthrough.

They seem to be using a lot of chip surface to do some passive components (big featureless area) that are needed for the analog part. Going to 5nm will make the chip more expensive.

The 5 micro-watts of power come from RF harvesting where they are constrained by law to only emit less than 2W. Bringing the emitter closer, they probably can afford a bigger power budget. It would be great if they could 10x the sampling frequency.

They mention the goal of using cheap manufacturing techniques but it seems like all those USB connectors make the thing really awkward. Does give it a cool Borg vibe, though.

They kept muttering science and research but video didn't show anything meaningful to 'help' science with this dress.

The motor cortex is the easiest thing to read via eeg that is consciously accessible. Any time you move or think of moving, it's activated. It's very easy to distinguish left side activation from right side activation, but much harder to determine whether you moved your left arm or leg. They use finger as a demonstration of sensitivity, but unless something has radically changed, finger discrimination from outside the skull is probably impossible, which is probably the reason it's a dress and not, say, a robotic exoskeleton.

The meaningful technological innovation which might be meaningfully exploited for science is the low-power sensor array.

Reply on news.ycombinator.com