As any cat owner will tell you, a cat’s ears are great indicators of its state of mind: pointed forward if they want your attention, turned backwards if they’re angry, and folded down flat when they’re afraid. Humans sometimes don cat ear headbands as a fashion statement, but sitting motionless those ears are more likely to confuse a cat than to provide any meaningful communication.
[Jazz DiMauro] aims to fill that gap by designing a cat ear headband that actually responds to your emotions. It does so by continuously taking an EEG measurement and extracting the “attention” and “meditation” variables from it. Those values are then applied to a set of servos that allow two-axis motion on each 3D printed ear. The EEG readout device is an off-the-shelf MindWave headset, which outputs its sensor data through Bluetooth. An Arduino then reads out the data and drives the servos.
Turning all this into a usable wearable device was a project on its own: [Jazz] went through several iterations to find a suitable power source and wiring strategy until they settled on a pair of lithium-polymer batteries and a single flat cable. The end result looks comfortable enough to wear, and the ears’ motion looks smooth and natural. All that’s left is to test it with real cats, to find out if they can now finally understand their human’s emotions too.
We’ve featured a few moving cat ear headbands before: one that moves along with your head’s motions, and another one with manual control. Today’s EEG-powered one shows yet another application for EEGs, which have been used for anything from invoking lucid dreaming to playing beer pong.
Interesting read. Almost the same design as Neco Mimi ( https://necomimi.shop/en ).
Can you really detect “emotions” with 3 EEG electrodes? I mean there is functional MRI at several orders of magnitude more complex, but I really need some input from someone actually into medicine. Seems pop-sci to me, no offense to the creators! I certainly like the idea.
Without trying to be rude, thats basically what NeuroSky MindWave is, in the best possible scenario you still only get a really rough estimate of emotions or ‘states’, you really need some medical professional to make actual sense of the data and it kind of requires you to train the device/software to detect your specific brainwaves first to even get those ‘maybe correct’ emotion/state detections. And then its still wrong half the time because yea, not enough electrodes.
A few years ago i sat on the sideline of a project for a car expo where somebody would get a VR headset and a MindWave headset, then they’d go a virtual lap across a circuit in a Porsche (so in VR) and the idea was that the MindWave headset would give us data that we could then use for a ‘live brain display’ to show how other people at the expo stand how excited the person in VR was from going across a circuit in a Porsche.
After ~2 weeks of lots of flaky/useless results from the MindWave device on several colleagues and test subjects it was decided to just fake the whole thing, people would still be given the MindWave device to wear, but instead they just made the ‘live brain display’ semi random and timed to the ‘exciting’ bits of the VR experience… lol.
ahaha!!! same here. I was supposed to use that thing for a video exhibition at the request of the artist, results were so inconsistents i had to fake it. It worked better :D !!!
..and then another marketing person / artist sees those demos and wants the same for their next expo / art installation and their techie after weeks of failure ends up faking it, and then … makes me wonder if there ever were any non-faked MindWave demos :)
A curious question is how the accuracy changes with increasing electrode count.
3 electrode does sound very low.
But 5000 might be impractical to implement in practice, not to mention potentially boarder into diminishing returns.
But surely it can’t be too much extra work to go to 5 (might still be too few), or perhaps 10.
it´s not the accuracy that changes by increasing electrodes count. It´s the number of channels. Each location on the skull taps some different signals. Increasing the electrodes just augment the resolution, but changes nothing to the accuracy. The accuracy depends on skin resistance, individual …
This is not even three electrodes, it´s two electrodes plus one reference electrode.
The only thing that this toy can assess is “neurocortical / brain neuronal activity is detectable on the skin surface of the skull in the form of electrical signals of millivolt amplitude, and the signal changes with the position of the electrodes, with the individual, and also with its mood .”
Finding a correlation that is reproducible and works with any individual is a marketing step that MindWave did, but their device does not work. It yes picks up signals. that´s it.
What I mean with “accuracy” is the accuracy of the final interpretation of the data, not the accuracy of the channels.
But yes, more electrodes is indeed just more channels for us to analyze. And potentially build a better picture from. Though, since both the individual (Humans aren’t identical copies of each other after all) and electrode position matters a lot in how to interpret the data gathered, then this more or less means that any EEG based controller would require user calibration.
Kudos for the technical R&D, but I suggest that if you want people to know how you feel, the human voice is the only technology you need.
Cool project! Shame the headband seems like a bit of a sham (or over-promise, under-deliver). That said you could probably get some similar data from heart rate and respiration rate, maybe throw in galvanic skin response if you really wanna go old school mind reading, so there are ways to build it without the pricy headset if you so desire.
I prefer not to let people know my emotions, I get hit less that way.
Maybe someone will make one that maps to muscle movements next.
Could map it to the eyebrows maybe.
Kind of a deep cut, this happened 10+ years ago: https://gizmodo.com/how-to-make-your-own-brainwave-controlled-cat-ears-5858995
What changed? Why post about it today?
This is only years old but thanks I guess?
On the other hand, Benjamin Choi, claims 95% accuracy using an algorithm with 23,000 lines of code: https://www.smithsonianmag.com/innovation/this-high-schooler-invented-a-low-cost-mind-controlled-prosthetic-arm-180979984/
There is a link to an Instructable, but almost no info on the mind control option.
I have a Mindflex game, a few years earlier than the Mindwave 2 and I’ve read that it uses NeuroSky parts. After trying it on me and a 6 year old, I feel it is definitely better than chance, maybe 60% to 65%.
Does anyone have an expert opinion on this application?