A fresh talk on the current state of #PsyLink, the open source myoelectric neural interface:
A fresh talk on the current state of #PsyLink, the open source myoelectric neural interface:
Our next open chaos talk is planned for tomorrow, at 20:00 in our space in cologne ehrenfeld. yuni will talk about PsyLink, an open source project about measuring electric signals from the skin to use them as computer inputs.
More details on our blog!
https://koeln.ccc.de/updates/2025-01-23_OC_PsyLink_neural_interface.xml
The #PsyLink UI can now automatically press a key for you if it detects gestures!
So finally it's possible to play games with it! 🥳
Will record a video tomorrow =)
Yay! Finally got the gesture prediction to work in the #PsyLink UI #rustlang rewrite!
First you have to do a calibration step where you do various movements as instructed on the screen. It trains a small neural network on that data with https://burn.dev
Then, it can recognize the gesture you're currently doing based on new incoming signals.
In the screenshot on the right, you can see the prediction results: "0" (= arm at rest), and when I performed a certain gesture, it changed to "1".
Using https://burn.dev for pure #Rust #MachineLearning on #PsyLink datasets:
https://codeberg.org/psylink/psylink-datasets
Check them out if you're researching #BCI / #EMG, they come with videos and detailed documentation.
Currently working on integrating this into the graphical user interface. Will be tiny enough to run 100% offline on your own device (hopefully even phone!) in the time frame of minutes 🥰
PsyLinks electrode armbands assembled!
If you're interested in #neurotech, #EMG, #BCI, let me know and I can send you one :)
Find out more at https://psylink.me
I'm also working on an #Android app for displaying the #EMG signals of #PsyLink and for doing magical gesture control of the phone.
It's written in pure #Rust and works on #Linux already, but on android, #bluetooth doesn't work yet. I fear I will have to dabble in #Java to fix this. 😰
Oh the agony. But can't wait for it to work eventually =)
Thanks to @PrototypeFund for funding this!
The #PsyLink assembly time has gone down to 1.5h since I'm using partially pre-assembled PCBs 😍
Still too long. Should get even faster once I find out how to remove all through-hole components - adding these is the bulk of the work since I have to manually trim them to avoid skin scratches, and then electrically isolate the bottom side.
#electronics #arduino #ble #bluetooth #neuralinterface #emg #eeg #bci #neuralink #diy #pcb