Author Topic: Seeking glove with knuckle-bend sensors  (Read 1759 times)

0 Members and 1 Guest are viewing this topic.

Offline wolfv

  • Thread Starter
  • Posts: 269
Seeking glove with knuckle-bend sensors
« on: Thu, 26 July 2018, 00:12:37 »
I am in the preliminary stages of designing a radical keyboard.
Like the DataHand, the user's hand will not move, only the fingers will move a very small distance.
And like the DataHand, the keyboard layout will have 4 layers.

User input will be finger angles formed by knuckles.
I am considering two different ways to implement a physical keyboard that can sense the knuckle angles.
The next two sections describe the two physical keyboards:
  • Sensor Glove
  • Finger Cage
Please tell me of potential problems I may have overlooked.
I would much rather know about problems now, than after building it.

Sensor Glove
Here is an example of a glove with sensors:

Fabric bend sensors are located above each knuckle with bend accuracy and repeatability at +/- 1.5 degrees.
Haptic feedback is built into the glove.  More details at:
https://www.bebopsensors.com/bebop-sensors-announces-forte-data-glove-arvr-applications-environments-gaming/
https://www.youtube.com/watch?v=fM1UTwnIn6Q

A sensor on each knuckle would drive the keyboard firmware.
The finger joints are named:
  • the metacarpophalangeal (MP) joint
  • the interphalangeal (IP) joint
The ring finger in this diagram has the two joints labeled:

Thus there are 4 virtual keys per finger, which are pressed by:
  • extending MP joint
  • flexing MP joint
  • extending IP joint
  • flexing IP joint
I did not find a sensor glove in production.
I have requested a Data Glove and development kit from BeBop Sensors but have not heard back from them.
Do you know of a glove with accurate sensors on the MP and IP joints?

Finger Cage
If a suitable sensor glove is not available, I could make a "finger cage".
A finger cage mounts buttons around the user's fingers.
A user inserts their fingers into the cage and the buttons sense the user's finger-joint movements indirectly.

The ring finger in the above diagram is annotated with 4 schematic buttons.
This is where the buttons would be mounted in the finger cage:
  • 3 buttons around the fingertip similar to the DataHand finger well, but without the side buttons
  • 1 button on top is labeled "knuckle button" in the diagram
A finger cage would have 4 such buttons on each finger.

Here is how the buttons would sense joint movement:
  • Moving the MP joint presses the "knuckle button" or "tip button"
  • Moving the IP joint presses the "nail button" or "pad button"
The prototype finger-cage buttons would be OMRON VX-01-1A2 limit switches, which have good tactile snap and a maximum operating force of only 0.25N {25 gf}.
Fingers would press directly on the red plunger (no keycaps):


The "knuckle button" is an untested idea as far as I know (please tell me if you know of such a thing).

The finger cage has some disadvantages compared to a glove. A user:
  • can not simply lift fingers off the keyboard when not typing (e.g. pause to think), must hold fingers in the neutral position or pull out
  • must be careful not to accidentally hit a button when moving fingers in and out of the cage
  • needs to use mouse-move keys similar to DataHand (a glove would use hand gestures instead of a mouse)
Do you think a finger cage with knuckle buttons could be a practical keyboard?

Thank you.
« Last Edit: Thu, 26 July 2018, 10:05:13 by wolfv »

Offline MatchstickMan

  • Posts: 131
  • Location: Seattle
  • CAD Jockey - YOLO Enthusiast
Re: Seeking glove with knuckle-bend sensors
« Reply #1 on: Thu, 26 July 2018, 09:07:41 »
The finger cage has some disadvantages compared to a glove. A user:
  • can not simply lift fingers off the keyboard when not typing (e.g. pause to think), must hold fingers in the neutral position or pull out
  • must be careful not to accidentally hit a button when moving fingers in and out of the cage
  • needs to use mouse-move keys similar to DataHand (a glove would use hand gestures instead of a mouse)
  • a neutral IP-joint angle of 45 degrees may not work well (a glove can sense any joint angles)
  • a neutral IP-joint angle of 90 degrees may not be ergonomic

I feel like a glove would have a lot of the same disadvantages that a hand cage would.
  • An inadvertent twitch of your finger could cause a "keystroke"
  • You're always in "typing" mode as the glove would be looking for input, unless it has position awareness, I guess
As for the sensors themselves, have you considered a series of load cells that measure the strain of the glove "joint"? Calibration might be difficult with different hand sizes and whatnot, but that should allow you to get some solid information on finger position through its movement. But I think having a load cell at the base (hand-side) of each MP and IP joint and translating the strain to movement could be a really interesting solution.

All that said, I imagine the Data Glove has a much more elegant solution haha

I'm excited to see what you come up with!
Current Projects:
- SouthPad -- Left-handed Numpad
- Full Split (Currently on hold)

Offline wolfv

  • Thread Starter
  • Posts: 269
Re: Seeking glove with knuckle-bend sensors
« Reply #2 on: Thu, 26 July 2018, 11:32:02 »
I feel like a glove would have a lot of the same disadvantages that a hand cage would.
  • An inadvertent twitch of your finger could cause a "keystroke"
  • You're always in "typing" mode as the glove would be looking for input, unless it has position awareness, I guess
Hi MatchstickMan,
Thanks for the information, I did not think of that.
An inadvertent twitch of your finger is a disadvantage of a glove because no resistance.
DataHand and finger cage switches provide some resistance.
So a finger cage might be better than a data glove.

As for the sensors themselves, have you considered a series of load cells that measure the strain of the glove "joint"? Calibration might be difficult with different hand sizes and whatnot, but that should allow you to get some solid information on finger position through its movement. But I think having a load cell at the base (hand-side) of each MP and IP joint and translating the strain to movement could be a really interesting solution.

All that said, I imagine the Data Glove has a much more elegant solution haha

I'm excited to see what you come up with!

I was looking at this Piezoelectric film sensor:
 http://www.te.com/commerce/DocumentDelivery/DDEController?Action=showdoc&DocId=Data+Sheet%7FDT_Series_with_Riveted_Leads%7FA1%7Fpdf%7FEnglish%7FENG_DS_DT_Series_with_Riveted_Leads_A1.pdf%7FCAT-PFS0005
But developing a sensor glove is too much project for me.
A finger cage would be easier to develop.
« Last Edit: Thu, 26 July 2018, 11:34:12 by wolfv »

Offline wolfv

  • Thread Starter
  • Posts: 269
Re: Seeking glove with knuckle-bend sensors
« Reply #3 on: Thu, 26 July 2018, 13:04:06 »
There is one experimental glove that provides haptic resistance:
https://ucsdnews.ucsd.edu/pressrelease/a_glove_powered_by_soft_robotics_to_interact_with_virtual_reality_environme
Very clever :thumb:.

But it will be a long time before something like that is in production.