Six Bone Gloves - Concept

ComradeZoned

Hello all,

My dream for the Chordata project is to enable non-invasive finger tracking for the existing motion platform. A previous discussion exists pertaining to a 16-bone model but has likely been abandoned. I figured this is not only feasible as is, but could be done with much less components, at over 140 US dollars in reduction of costs, per pair of gloves.

There are two different boards in this solution, the Opisthenar (back of hand), and the Manipulator (the fingertips).

The idea is to mount these components to a glove with minimal reduction of dexterity.
The Manipulator boards have a total of 8 components onboard to reduce size: C1-C5, a pull up resistor, and a LSM9DS1 IMU. These "manipulators" would be connected through JST-SH-04 (qwiik) twisted pair, straight-through (not crossover) cables, and cannot function without the Opisthenar main board.

All EEPROM and other components are placed on the Opisthenar board for consolidation of power and data lanes. This hand based board has six connectors for Index, Middle, Ring, Pinkie, Right Thumb and Left Thumbs, allowing the same motherboard to be used for either hand.

A single 3.3v LDO (low dropout regulator) is used for power correction (making this system functional up to _5.5 volt input). I've also added a schottky rectifier diode to prevent reverse polarity. Both of these components can be bypassed if deemed unnecessary by the end user. Also included is a single IMU to replace the hand's K-Ceptor unit.

The current schematics are available for download on EasyEDA, links below.
Both schematics are "panelized" by default. A minimal quantity order of 5 through JLCPCB will produce 10 total Opisthenar (1x2x5) and 200 total Manipulator boards (5x8x5).
Opisthenar - Hand motherboard
Manipulator - Finger tracking modules

Github homepage, Currently Empty

I have ordered a set of PCBs and will try to keep this feed alive with progress once I start building. I plan on making three total Mocap suits using this solution + the base Chordata platform.

All constructive criticism is welcome - this is my first project of the sort, and I'm sure I've bodged something here.

daylanKifky

🤩 🤩 🤩
This is looking great!

A shout out to you for sharing your work from the beginning! I also like the fact that you gave names to the boards 😉

ComradeZoned All constructive criticism is welcome - this is my first project of the sort, and I'm sure I've bodged something here.

The only detail I would point out is the multiple EEPROMs, but is not your mistake. Only one should be enough to store the calibration data of all the fingers, BUT! it would require to modify the notochord software to identify the Opisthenar as a special case.
Perhaps it can be modified in the future, once the planned mayor rewrite of the notochord is done.

Flavia

ComradeZoned Hi there! I'm Flavia, the Communications Lead of Chordata Motion. Congratulations on the amazing work you're doing!!!! Yesterday I did a post on our social media channels talking about your work. I hope you like it!

ComradeZoned

Dodo2207 how you can get the exact bending of each finger?

So this portion of the project is not completed, it's still a living concept at this time.

I plan to experiment, but the current idea is to approximate the rotation of the rest of the joints by comparing the distal phalange's forward vector to the forward vector of the hand bone. Using the tracking data for the accelerometer would assist in resolving orientations.
For example, 270 degrees relative distal rotation would indicate the joints are likely each around 90 degrees, while the accelerometer would tell how much the bone has moved to get to that location.

Adding one more tracker per finger on the proximal phalanx would resolve the issue in the situation you propose otherwise - I'll consider designing another version, but the costs of such a system is inherently higher.

I also have not decided which resistor value(s) to use to set the address for the I2C switch. daylanKifky Any recommendations?

Dodo2207

@ComradeZoned, your fingersonsors sounds amaizing 👍
I will definetly try these after finishing and experimenting with the mini Kcs and mini hub which are in work.

Question about your sensors:

If you only have 5 Manipulators and 1 Opisthenar, on each hand, how you can get the exact bending of each finger?

If I move the metarcarpels bone and dont bend the phalanges bones wouldnt I get the same result as I bend the phalanges bones as well?

Im exided for your first tests and thanks for shareing 👍

daylanKifky

ComradeZoned I also have not decided which resistor value(s) to use to set the address for the I2C switch

Take a look at the LTC4316's datasheet, there's a table with the recommended resistor values. That table is partially copied in our wiki: https://wiki.chordata.cc/wiki/ID_Module#Soldering_the_translation_value_resistors

daylanKifky

oops, I probably got the question wrong. If you are talking about the switch in the hub we use 10K but also lower values should work

ComradeZoned

daylanKifky I was referring to the switches I have placed on the 'opisthenar' board. Just trying to make sure I don't have address conflicts/have the theory of that chipset wrong.
If the current setup is not possible, I will redesign with discreet address translation.

daylanKifky

Hi!

In order for the PCA9548 in the Opisthenar not to clash with the address on the central Hub you have to assign a different address to it. By setting A0, A1 and A2 differently.
Right now they are configured as in the main hub, giving you the address 0x73.

You can, for example, pull the MSB (A2) to VCC to obtain 0x77.
Only one branch from the central hub is open at any given time, so both hands won't clash between each other.

Let me know If I answered your question this time.

PS: In this project the complicated names for the different components are already the norm, but where does Opisthenar come from?

ComradeZoned

This is good to hear, thank you for clarifying. I pretty much jumped on this hoping that would be the case.

daylanKifky where does Opisthenar come from?

It's the anatomical term for the back of the hand c:

ComradeZoned

I've released revision 3.4 to correct the pull-up state for MPLX A2.

I've already ordered PCBs for rev 3.3, but can correct this issue for now with a knife and a "bodge wire".

ComradeZoned

Turns out it's a pain to add bodge resistors directly onto TSSOP packages. (For previously resolved issue.) Nothing some thinner gauge wire and some glue can't handle. Hopefully I can still work it out close to NASA specs - just for fun.

daylanKifky

ComradeZoned Turns out it's a pain to add bodge resistors directly onto TSSOP packages.

😱
At least you just have to do it for only 2 boards

Those NASA specs are quite interesting, where did you find them?

ComradeZoned

daylanKifky The official pictorial reference may be found here: https://workmanship.nasa.gov/insp.html

ComradeZoned

Taking a break, it's vacation time around here. Project is not abandoned until I say so.

zaanen

ComradeZoned maybe a stupid question i don't know noob at pcb production but is rigid-flex pcb bord not an outcome for the gloves? because the flexible part would go with the bending of the fingers and you hold flexability in finger movement while IMU ar on them ?

Maybe it is just idiot thinking

zaanen

would love to see this working for hand tracking movement in racketsports. will follow your progression closely

ComradeZoned

zaanen Not a dumb question. The main problem with IMU's is that they are highly susceptible to torque. Any slight bend in the IMU or the substrate it is placed on will cause deflection in the tracking.
A flex-PCB would be a good interconnect, but placing the components on a flexible PCB is recipe for disaster.

I hope this answers your questions.

ComradeZoned

If I have learned one thing on this project, it's that I abhor crimping .4 mm pitch JST connections 😅 I may come out with a better concept sometime when I return to the finger project.

Main priority right now is getting my other units built - all of them are 95-ish percent completed. No promises I'm afraid.