A Lazy Programmer’s Datagloves
Arne has commissioned us to make him a pair of datagloves for coding. He wants to be able to code with the slightest movement of his fingers. We’re working on a series of test gloves to evaluate different sensors for detecting slight finger movement.
Arne has been thinking this idea through in quite some detail. The goal is to be able to input all the keys of a keyboard with as little finger movement (effort) as possible. Instead of only sensing downward movement of the fingers, the sensors on the glove should also detect upwards movements. Meaning each finger can move either up or down – allowing it to “input” two different states.
Links:
Hannah’s Flickr set: https://www.flickr.com/photos/plusea/albums/72157664864833147
Mika’s Flickr set: https://www.flickr.com/photos/14412219@N04/albums/72157694775448464
Code: https://github.com/KOBAKANT/KOBA/tree/master/COMMISSIONS/10-Arne_LazyGloves
E-Textile Dataglove Overview: http://www.kobakant.at/DIY/?p=6730
Dataglove Overview: http://www.kobakant.at/DIY/?p=7114
DIY E-Textile VR Glove: http://www.kobakant.at/DIY/?p=7190
Technical documentation:
Construct a tailored dataglove pattern: http://www.kobakant.at/DIY/?p=7203
Sensitive Joints – EMF sensing of the hand:
First Meeting (March 17th?)
In a frist meeting we discussed Arne’s idea and decided it is something we can make, but would start with a series of tests on what type of sensor to use. Sensors to test included:
– textile pressure/bend sensors
– DIY pressure/bend sensors (made on kapton)
– commercial bend sensors (flexpoint)
– analog magnetic field sensors and neodymium magnets
Analog Hall-Effekt-Sensors:
SS39ET/SS49E/SS59ET, 1,5 mA 2,7 → 6,5 V dc
https://de.rs-online.com/web/p/hall-effekt-sensoren/8223769/
https://docs-emea.rs-online.com/webdocs/13e7/0900766b813e7970.pdf
SS490 Series (SS495A1, SS495A)
https://docs-emea.rs-online.com/webdocs/002e/0900766b8002e131.pdf
SS490 SS496A1
https://docs-emea.rs-online.com/webdocs/0445/0900766b80445225.pdf
Sensors…
I ordered samples of these magnetic field sensors:
Hall-Effekt-Sensor Analog, 1,5 mA 2,7 → 6,5 V dc
RS Best.-Nr. 822-3769
Marke Honeywell
Herst. Teile-Nr. SS39ET
https://de.rs-online.com/web/p/products/8223769/
Hall-Effekt-Sensor Ratiometrisch, 1,5 mA 2,7 → 6,5 V dc
RS Best.-Nr. 822-3787
Marke Honeywell
Herst. Teile-Nr. SS49E
https://de.rs-online.com/web/p/products/8223787/
Hall-Effekt-Sensor Ratiometrisch, 1,5 mA 4,5 → 10,5 V dc
RS Best.-Nr. 479-6878
Marke Honeywell
Herst. Teile-Nr. SS496A1
https://de.rs-online.com/web/p/products/4796878/
Arne trying on the first full EMF prototype:
Arne trys out the final prototype:
Demo of how the magnetic lever works:
See Yourself Sensing: Madeline Schwartzman
1:36:37
EMF sensor prototype using kapton arch to TRANSLATE bend into distance:
Ideas for solving it all with magnetic field sensors
First prototypes to test different sensors
Placement of Analog Hall-Effekt-Sensors on my hand:
Notes from second meeting where various sensors were tested
Video of Arne’s desired minimal finger movements:
Stills from video:
THUMB: up, down, neutral, left, right
FINGERS: up, neutral, down
Commercial flex sensors:
Textile pressure(bend/stretch) sensors:
Magnetic sensors:
Based on the testing with Arne of the above, we decided the first full (one hand) prototype will consist of:
thumb: 2 bend sensors + 2 (or more) magnetic field sensors
index finger: 1 proximal knuckle bend sensor
middle finger: 1 proximal knuckle bend sensor
ring finger: 2 bend sensors (proximal knuckle, middle knuckle)
little finger: 3 bend sensors (proximal knuckle, middle knuckle, distal knuckle)
Pattern
I wonder if we can get the sensors to work only on the hand knuckles, if we could use a minimal pattern like this:
French: Fourchette, the French word for fork. In English, a technical term for a type of dessert fork, a component of a glove.
>> https://www.inesgloves.com/products/fingerless-half-scoop-gloves?variant=3007447300
Or try this pattern:
>> https://ohicosplay.tumblr.com/post/125019474707/adreecosplays-adreecosplays-skyes
First EMF Prototype for Betatesting
Hardware
Teensy Low Cost (32 Bit, 11 analog) with ARM Cortex-M0+ processor:
>> https://www.pjrc.com/teensy/teensyLC.html
>> https://www.exp-tech.de/plattformen/sonstige/6349/teensy-lc
HCF4051 Single 8-channel analog multiplexer/demultiplexer on each hand to capture all the bend sensors. Magnetic sensors go directly to board.
>> http://www.st.com/resource/en/datasheet/hcf4051.pdf
Connectors…
Could be interesting to get a magnetic quick break-away connector for the USB connection to Teensy on wrist.
CODE
Arduino Code
HCF4051 mutliplexer channel parsing code >> https://playground.arduino.cc/learning/4051
Colours???
mud pink, light blue, navy blue, yellow, (silver), (dark grey), black:
Code and visualization:
Arduino to Processing graphing multiple sensors
https://github.com/ericjforman/GraphMultiple_p3/blob/master/GraphMultiple_p3.pde
Using Teensy LC:
NOTES
Other sensors…
ANALOG Magnetic Sensors
analog Hall magnetic sensor module
https://tkkrlab.nl/wiki/Arduino_KY-035_Class_Bihor_magnetic_sensor
KY-035 ANALOG (KY-003 DIGITAL)
http://arduinosensors.nl/index.php?route=product/product&product_id=72
AMR Sensors (Magnetic Sensors)
These products are magnetic sensors using AMR elements whose resistance varies according to the strength or direction of the magnetic field. Combination with a magnet permits non-contact rotation detection or position detection.
https://www.murata.com/en-us/products/sensor/amr/position-rotation
Adafruit: Hall effect sensor – US5881LUA
High Sensitivity Bipolar Latching Digital Hall-Effect Sensor ICs SS360NT/SS360ST/SS460S
https://docs-emea.rs-online.com/webdocs/13e7/0900766b813e7973.pdf
First prototype feedback
Sensors moving around and small movements come from fabric, not actual finger bend.
http://minimal.uk.com/crossfit-gym/barehand-gloves-by-raww.html
Visit (11/5/2018)
Arne came with feedback and decided to work on a next iteration that explores the use of non-stretch fabric/material to maintain fixed distances between sensors to (ideally) eliminate cross-talk. Are we getting to something like the P5 Dataglove for gaming?
Before Summer (20/6/2018)
Before leaving for a long summer holiday, we made this quick test to see how it could work to use non-stretch fabric in Arne’s glove design.
It looks promising and we will get back to it when we get back to it.
New Ideas (11/10/2018)
Five months later and a summer in between…. we pick up work on Arne’s programming gloves again. He has written to say that bending one finger pulls at the fabric and affects other sensors on the hand. In an attempt to solve this issues we thought about trying to construct part of the glove where the sensors are mounted out of a non-stretch fabric.
Here a video of idea sketching:
The downside of a non-stretch band around the palm is that it can’t be strapped as snug. We were just recently working on Tom’s limb link sensor and using a neoprene with one raw side for grip against the skin (the other side is fabric fused). We have some 4mm thick neoprene of this kind that looks like it will work quite nice as a more stable material for a palm strap – instead of just the powermesh glove previously. Here a video of the strap to show how little it moves when the fingers move. As soon as the thumb moves it begins to move more, even when the extra thumb strap is removed.
Instead of mounting a full glove with this palm strap, I’m experimenting with having the sensors fixed on the PALM BAND and inserted into little kapton SHEETHS mounted atop individual elastic FINGER RINGS. This looks to be working very nicely. I just wonder how precise the design needs to be and if what I am now making to work on my hand will also work as well for Arne’s hand.
FINGER RINGS
The loop on the ring for the sensor shaft to slide through is made by crimping a female snap on some soft stretch elastic to make a closed loop ring. Then a strip of kapton is folded and hole-punched and the male snap presses through the two holes and into the female snap on the ring to form the sheath and attach it to the ring. The nice thing about these snaps is that they work as a pivot around which the materials can move as the fingers move left and right.
PALM BAND
The sensor shafts are held on the PALM BAND with another snap, similar to the previous design.
Meeting and Test of PALM BAND and RINGS (12/10/2018)
QUESTIONS TO ASK:
Q: Any chance we could combine all kapton sensors into single piece and solve circuit in copper tape?
A: NO
Q: Is it a problem that rights shift over time?
A: No
Q: Does thumb movement affect finger sensors?
A: Not a problem
Q: Do magnets stick to sensors and solder too much?
A: Yes, but solved with weaker magnets
Q: Is the length of the levers okay?
A: OK
NOTES:
– Distance of sheath mounted on FINGER RING needs to be as low profile as possible (close to skin) while still allowing sensor lever to move back and forth smoothly (no resistance).
– Elastic solution for adjustable sensor lever to magnet lever should be replaced by solution that does not meet with skin.
– PALM BAND is tight and sweaty – make it 3cm longer and mount snap at next meeting, possibly perforate neoprene to make more breathable. Could try making PALM BAND as narrow as possible – only as wide as the snaps (7mm).
– Forget bend sensor between index and thumb for now, instead mount second thumb sensor.
– FINGER RING fitting: little, ring = good, middle, index, thumb = slightly tight.
– Solve circuitry more elegantly.
– Order larger diameter (4mm), lower profile (0.5 or 1mm), weaker magnets because current magnets (from Modulor) are so strong that they their attraction to the Magnetic Field sensor component messes with their free/smooth movement.
– Fix sensor lever to magnet lever at far end?
– Fix magnets more permanently (not immediate issue).
Discussed Ideas:
– To mount sensor on finger and have angle created
– To have finger bend move magnet back and forth (linear displacement) instead of angular displacement….. but came to conclusion that then the mounting of FINGER RINGS would need to be exact. Discarded this idea again for now.
PROTOTYPE FOR NEXT MEETING (early November):
– new FINGER RINGS with low profile sheath.
– longer/narrower/more breathable PALM BAND.
– MAGNETS.
– second thumb sensor (1.6cm apart).
– tidier wiring.
References to other work:
DIY bend sensors just like in the Dataglove Flex Sensor Rig.
>> http://www.kobakant.at/DIY/?p=7113
Breadboard the inputs with pull-up pots just like for the DIY VR Glove.
>> http://www.kobakant.at/DIY/?p=7190
Dataglove Flex Sensor Rig
>> http://www.kobakant.at/DIY/?p=7113
DIY VR Glove
>> https://www.flickr.com/photos/plusea/albums/72157679861886694
Maurin is using both analog ports of the Teensy to read even faster from two multiplexers.
https://matrix.etextile.org/
Updated Design (12/12/2018)
Demo:
Ring Version Explained:
Textile Version Explained: