Building a pattern recognition system

We have learned a lot from the first few versions of the assistive arm. Using what we have learned we are moving from using a single muscle sensor and signal to using a method that will require us to pick up multiple signals from the skin using dry electrodes. We need to figure out how to process those signals, recognize the signals and then use recognized signals as a control signal we can send to the actuator. Luckily, we have tested many ways and there are some commercial solutions out there and even some open solutions. We need help to put this all together.

If you have experience in electronics, myoelectric signals, pattern recognition... and can help us figure out how to get things working on a Raspberry Pi, then please help!

Requirements:

We are still scoping out the requirements - Help us :)

- We want to use up to 17 electrodes (8 pairs and 1 reference electrode).
- We need the full raw signal to be captured and transported as 8 separate channels to the Raspberry Pi.
- Some have suggested we have a minimal bit depth and sample frequency I.E at least 16-bit adc and 1000 s/s.
- The connector between the electrodes and Raspberry Pi needs to be small and easy for a child to attach and detach.
- The Raspberry Pi will be located on the hip, therefore the wire length should be about 60cm.
- Each channel (pair of electrodes) will need their own color wire so that assembly can be accurately done.

Possible solutions:

We tested things with a OYMotion GForce Pro for this, like this one: https://www.aliexpress.com/item/32965841641.html?spm=a2g0o.detail.1000060.1.101e38adwMZV75&gps-id=pcDetailBottomMoreThisSeller&scm=1007.13339.99734.0&scm_id=1007.13339.99734.0&scm-url=1007.13339.99734.0&pvid=5532a2e9-0531-482e-af36-225132a24e44

Another option is this open solution: OpenBCI kits: https://shop.openbci.com/collections/frontpage