Open Brain Computing Interface | Hacker News Reader

Open Brain Computing Interface

openbci.com

81 pointsbayesianhorse28 comments

People interested in DIY neuroscience might also be interested in the SpikerBox [1] and associated neuroscience course [2]. Although not in the same vein as the OpenBCI, it provides a hacker-friendly way to learn about principles in neuroscience and explore at your own pace.

[1] https://backyardbrains.com/products/spikerbox [2] http://mcb80x.org

It's mind blowingly expensive. A very quick breakdown of the main parts:

ADS1299 ~38USD for 1k+

ATMEGA328P ~ 2USD for 2k+

RFDuino ~ 20USD for 10 +

Accelerometer ~ 2USD for 1k+

Board ~ 5USD assembled for 1k+

I'm also not sure I'd want to have an RF module near a 24 bit ADC.

Total is probably around 75USD for 2k+. Simple layout as well.

Thank you! For a open source hardware project I was looking for the BOM(and wondering at it's real price). Presumably there is some R&D recoup baked in to this price, but when other micro controllers and kits sell for ~$25-40 to see this over a factor of 10 more expensive was surprising.

I was looking at mouser when you posted this and was getting fairly similar numbers (though not quite as low). The price certainly does seem high.

What sort of markup would you normally expect for relatively simple electronic projects like this?

If it's been done nicely 100% would seem fair to me for a low volume project like this. That's only my opinion however.

Thank you for the opinion. The price is the main reason I don't see myself buying one in the near future.

I rationalized the price because I figured it is very hard to build a nicely running prototype board with the (specialized) amplifier chip, and that there was at least a 100 or 200$ pure "convenience" charge in it.

For an EE it's pretty easy to put these parts together and make them 'work', but it's fair to say that making the most out of a 24bit ADC is no easy task. That is unless you want 8+bits of noise on the line. You need to take great care decoupling the digital side from the analogue side, and also with the routing of the ADC inputs - guarding / shielding etc. If that's all been done properly it would certainly be worth paying a premium for the time that's gone into the design. I guess the question is has it?

I don't know if it is. But I also don't believe most people interested in BCI or biofeedback are electrical engineers...

My entire lab is filled with EE's focusing in neuroscience and brain-machine interfaces. We use implantable electrodes though, much better SNR and fewer corrupting noise sources.

Noob question: When you say "implantable electrodes", you mean for humans, or for animals?

Usually humans. I'm not involved in this kind of research, but BCI with implanted electrodes probably refers to cases where these electrodes are implanted for some other reason.

For example it can be necessary to find an epileptic seizure hot spot because you can't stick someone in an MRI tube for a few days on end.

It's hard to actually make a case for BCI as an assistive technology, because every single sceletal muscle in your body provides a better input signal than EEG ever could, and there aren't so many cases where people don't have any voluntary muscle function left. (My opinion... need not be correct!)

An open source EEG hardware project has existed for years in openEEG, Olimex sells a surface-mount version with enclosure for $100 (though it is only two-bipolar channels) https://www.olimex.com/Products/EEG/OpenEEG/

I'm very interested in this hardware, but the price has been a deterrent for me as well. I would have ordered it already for $199. But for $449 + extra for the electrodes, etc., it's a pretty serious investment for a fairly unknown device.

My main interest in this tech is to be able to build a machine learning classifier for meditative states. Then when I have that classifier, I'd like to make a simple mobile app that lets people "record" and "measure" their meditation sessions.

You can do that with the Mindwave Mobile for around $150, and there are cheaper and more expensive versions available...

Too low resolution to do that sort of science. Mindwave Mobile is only a single channel.

The device seems to be designed for Neurofeedback, and it can measure frequencies relatively accurately. It has two built-in feedback measures "attention" (mindfulness meditation) and "meditation" (more like loving kindness meditation). The raw signal can be used to develop your own classifiers/feedback measures.

Currently I think the practicality of one dry sensor on a wireless headset, combined with the cheap price, beats every competitor on the consumer level.

I did some work with the Neurosky Mindwave Mobile headset. Compared to OpenBCI it is cheaper and more practical to use, but on the other hand it only has one channel and a resolution of 512hz. It is mainly useful for neurofeedback training and assisting learning technologies. It has been used for simple BCI tasks.

OpenBCI offers more channels, and I suspect a higher possible resolution. You'll have to get more involved with the hardware though, both in terms of rigging a device and electrodes, and programming the hardware on the chip (arduino). More channels means that you can increase spatial resolution, which allows for easier discrimination between mental "switches", more advanced diagnosis and feedback.

Still, keep in mind that you need to get dirty with the hardware!

sigh

I need help. I showed this to my son (14) as a prop to encourage interest in STEM work.

"You mean I could control Minecraft with my brain?"

Yes, son, that's exactly what it would be for, so you can look at a block and make it disappear.

Whatever are his motives, if he develops the technology and doesn't crap on humanity the way the patent system allows one to do, other people will see the potentials and make good use of it. That's why academic freedom is important :-).

In other word, let him make bubbles… cf https://gist.github.com/stoutbeard/4158578.

As long as he makes something, I'm not too picky. I have a music maker, a graphic novel maker, a musical theater performance maker[1], and...I don't know, yet. Of course I'd love him to be a software maker, but all I really want is his creativity and intellect to grow through making something he loves.

1. I was in theater in high school, but I never "made" a performance. The difference is, well, acting. :)

Way overpriced. Emotivs epoc was $300 and did 16 channels. Or 14... I forget.

It's also less flexible, and you can't use it as easily for EMG and ECG, which is a major plus in terms of medical prototyping, biofeedback, etc.

I just checked, it's 400 USD, raw data is another 300 USD.

No there is a firmware hack called emokit which allows data with the $300 headset (correction its $400 now LMAO). However, emotiv is by far the single worst company I ever dealt with. Worst product support experience of my life. Having bought one of those epoc headsets I have to say its the very definition of flimsy, a half baked buggy product with planned obsolescence and proprietary restriction built in on every level of design. Emotiv even censored their forum and deleted my username to hide how they totally fucked me over. I support the openbci guys they have done a great thing by making a open source EEG sensor kit but well they are being rather greedy in the marketing. A terrible idea because right now they need to build a developer community around the device. They should be giving these kits away.

https://github.com/openyou/emokit

sad to hear that! My epoc is on its way.

If you plan to use Emokit, please drop me a line (email's in my profile) if you have any issues. I'm no longer the maintainer, but I like to keep abreast of its usage, and want to make sure that it continues working for new devices. (For context, I'm the original reverse-engineer behind Emokit.)

Awesome, will do! (and yeah, I've got emokit starred already :) )

Reply on news.ycombinator.com