Affordable Wheel Based Refreshable Braille Display | Hacker News Reader

Affordable Wheel Based Refreshable Braille Display

jacquesmattheij.com

231 pointsjacquesm67 comments

As someone blind, my two pens are that noise, power consumption, and fragility are things that could be compromised to a large degree. If I've had a braille display at school and uni, it would've helped with many of my math issues. If it works and it is cheap, it would be a big step forward for many people. Power is not that expensive compared to the current displays, noise could be mitegated or just powered through, and important stuff could be kept in good order. Money are harder to spend.

Screen readers are perfect for plain text and gui navigation, but a multi dimensional object like an equation are easier to understand if you have your hands on it.

I wonder if one could combine a low-fidelity braille display (maybe with heat or vibration (piezoelectricity)) with touch sensitivity and context-sensitive tts in response to touching/reading the display.

I was curious about just temperature.

From this pdf[1] it seems fingers have 3-5 cold sensing points per square cm, with cold sensing points being about 10x as prevalent as warm sensing points.

So I was thinking about using a combination of a thermoelectric cooler[2] coupled to some thin metal rods, which have individually wound coils around them near the top. The rods would be mounted flush to allow touching all of them at once. The cooler would ensure the rods are normally cold, and the coils would provide induction heating.

Thus by changing which coils are activated, the different rods would be either cold or not, and due to the active cooling and relatively low thermal mass should change state quite quickly. At least that's the idea.

Though perhaps we just don't have enough sensitivity in our fingers to pick this up. And it wouldn't work well for a portable device due to power draw.

edit: Found this[4] paper where they try a temporal method instead of spatial method to use temperature to convey information.

[1]: https://web.as.uky.edu/Biology/faculty/cooper/bio350/Bio350%...

[2]: https://en.wikipedia.org/wiki/Thermoelectric_cooling

[3]: https://en.wikipedia.org/wiki/Induction_heating

[4]: https://www.mdpi.com/2306-5354/10/10/1156

I love unique interface ideas, and this one definitely counts.

What are your thoughts on how to account for the user's fingers heating up the interface rods? I'm imagining someone going back to read a previous character, and their fingers having essentially "reset" the character from body temp.

This was just off the top of my head, but I imagined the thin rods having low enough thermal mass that the active TEC cooling would not cause fingers to significantly heat up the rods.

That is, the TEC cooler should be able to relatively quickly lower the temperature after the finger is removed.

Again, off the top of my head idea so no idea if it would work or not, but doing a quick viability check should be relatively easy and cheap.

I have thought of a similar design, but little solenoids inject themselves into the cold path bridging the gap, the TEC would chill a huge thermal mass and then the individual rods could be chilled and heated rather rapidly.

Would it help to have some kind of a reverse plotter?

Imagine an XY gantry like a 3D printer has. The axis are free, so the device reads the coordinates as the user moves them. It in turn has a little servo motor in there moving the "stylus" in the Z direction. When trying to view an image, white would be mapped to "down" and black would be mapped to "up". This would make it possible to "view" simple 2D plots, or perhaps even equations.

User experience could be improved by using force-feedback motors on the XY gantry, essentially nudging the stylus to stay on the line.

Something like this should be somewhat doable for about $200-300, I reckon.

Hard to say. Information density matters and also the speed with which one can explore the graph. The limited use cases also might make it impractical.

Beautiful example of the innovation narrative that keeps CTOs of the largest companies awake at night. A (very) intelligent outsider, with pen and paper, computer, a 3D printer who lives where Aliexpress delivers. Meanders and bounces between theory, practice and sleep deprivation can come up with a disruptive PoC in a matter of weeks [0].

I would really like this to work, and/or that it inspires others to make these braille readers affordable.

[0]: the inspiration for this project started 48 days ago: https://news.ycombinator.com/item?id=39159476

Not just any old guy but my good friend and our very own Jacques

https://news.ycombinator.com/user?id=jacquesm

I love how "science fair" this writeup is.

Personally, I would have looked at more options before going for the (rather large) motorized wheel route. Two that come to mind are reusing the ball-type powered typewriters from the 80s/90s since the balls had embossed letters and high-precision positioning already. A second is microfluidic displays, which cellphone makers toyed with for on-screen keyboard tactile feedback back in the early 2010s. And indeed a quick search shows that a University of Michigan team used exactly that for a braille display 8 years ago [1] and it's spinning out into a company now. The company that was working on this for "pop-up" touchscreens 10 years ago was Tactus [2]. On the electromechanical side, it looks like there's an open source movement already with some interesting results so far [3].

[1] https://www.youtube.com/watch?v=0fIg4rI4cDw

[2] https://www.youtube.com/watch?v=JelhR2iPuw0

[3] https://www.youtube.com/watch?v=BXi1tG78AW4

Tiny pinpoint holes, blow air through them. Could you adjust the size of the holes and the flow of the air so you could feel them distinctly? If so, they could be backed by much larger, far away, valves without the mechanical parts needing to be so fine and precise.

You only need to feel something, there doesn't need to be anything there - is it possible to have a voltage or capacitance charge or signal at a point which feels like a presence? "A smooth surface which feels lumpy" seems intuitively unlikely but stranger things have happened.

Yes, I was wondering if something could be done using 3d printed fluidic logic. If so then the entire thing could have no solid moving parts, and just be one big 3d print, except: (maybe) a single big rubber membrane in which dots are raised by fluid pressure. Plus of course the fluid pressure source; and the valves to interface the device to electronics, to input the display data. Otherwise it could be a big shift register in fluidic logic with an amplifier at each bit to output to each dot.

The difficulty with fluidic logic is that the devices from the original flowering of the field don't work at the low Reynolds numbers found in smaller devices. Maybe something from the new micro fluidic field would work, but I don't know if they can control sufficient pressure to be felt, or to raise a rubber membrane.

"A smooth surface which feels lumpy" seems intuitively unlikely but stranger things have happened.

I wonder how well it would work if you did something similar to Force Touch, but instead of one uniform trackpad it were separated into a matrix of small dots.

Yes, you can do that but you still need a cheap valve with a size ideally smaller than one dowel pin 'cell' at standard braille densities. AFAIK this does not exist commercially, if it does it would be cost prohibitive, and therefore would need to be designed in some unknown manner focused on low fabrication cost. Since most valves operate based upon electromagnetism (solenoid-driven) you also have an electrical control problem. All in all, not an attractive solution path.

The reason for suggesting that is because it wouldn't need tiny expensive custom delicate valves; connect the pinhole to a funnel which gets wider like the horns on this siren: https://stall.net/victorysiren/photo-pages/cars2/images/5SG-...

then you can use whatever large, cheap, widely spaced, easy to access, easy to maintain, valves you can find. The trade would be size, noise, and power, but the gain would be easy to build, mass produced parts, easy to repair, large and solid and relatively reliable.

Most valves are actuated by solenoids which use a lot of power. An array of them is not only physically huge but will exceed USB power availability at a count of one or two valves and you need many per braille block. Worse, you generally need to actuate huge numbers of them simultaneously. This approach is therefore difficult to consider for a real world deployment.

The top comment on this thread by gostsamo is "As someone blind, my two pens [pence?] are that noise, power consumption, and fragility are things that could be compromised to a large degree."

How many do you consider 'huge numbers'? There's another comment saying a Braille cell is 8 dots, and if you average changing half of the dots each time to change state, you could barely get into double digits to keep ahead of the fingers.

> "This approach is therefore difficult to consider for a real world deployment."

It only had 10 seconds of thought; I just skim read the article which was all about mechanical actuation and the difficulty of making so many small and precise and reliable actuators cheaply enough, and wondered about options that didn't require that part. Air, electric current (e.g. the vibration felt in a pot on an induction hob vibrating some loose rods without having to actually lift them), or a one-finger glove where there is only a few actuators and they change as the finger slides over a table, or no actuators - a design like the way a CRT screen has a scanning electron beam activating each pixel, although I have no ideas how a design like that could activate Braille dots.

I think PCB based designs will have a better Design for Mass Manufacturability.

See the work of Carl Bugeja.

https://www.youtube.com/watch?v=oa6sP-joAr8

Either motors, or solenoids, electromechanical brakes, compliant mechanisms.

This is the approach I used for my prototype (thanks to Carl!) and it does function.

However, since the area is quite "edge case" (rarely developed in public), there remain challenges in balancing the electrical and control requirements with physical requirements, electromagnetic density, part selection, manufacturing process and cost.

At that point it might make more sense to go back to a more traditional approach. Place each "dot" on a little magnet, and use a PCB with coils in it to actuate them.

A while back I made a (prototype) braille display with a similar concept but using linear sliders (per column) rather than a rotating wheel. Unfortunately I got side-tracked from the project and never figured out the best way to add actuators.

I was proud of the fact that my design could be laser cut from a single sheet and assembled with no glue or fasteners (well, minus whatever mechanism would be needed to actuate the sliders)

Image: https://retr0.id/media/38116918-4023-437b-9a48-d2ffb1d02dbf/...

Short demo video: https://twitter.com/David3141593/status/1639261097252233220 (in the video caption I noted high friction, it was totally fine after sanding)

Another idea: Instead of actual tactile dots, use a voltage between two appropriately sized/positioned contact points, high enough to be felt by the fingertip, but not so high as to be uncomfortable. Then you just manufacture it as a regular PCB.

(It would probably be required to multiplex the voltage between different dots, so that the finger current always flows within the small area of individual dots, rather than between dots.)

It might work, and it's an incredibly simple solution, but I'd want to check out if there is any chance of long term nerve damage. The user will be using it many hours a day nearly every day.

This was tried before in haptics research, and the participants nerve systems quickly adapted to suppress initial threshold stimulus.

i.e. one fatigues the sodium channel properties rather quickly.

There is also the emotional scars of working at a Screaming Monkey Medical Research Center... the burnt hair smell never really washes off. lol ;-)

Only partially joking, we repaired one of the old braille displays which used 400v piezoelectric actuators about 4mm away from peoples fingers.

Best regards, =)

Reading through this, it really strikes me that you're using the wrong printer for the job.

This sort of component is a natural fit for a resin printer. The registration slits from #8 can be created reliably with a resin printer, and I'm confident you'll get better dots as well.

Furthermore, the speed of printing is by Z axis for a resin printer, not volume of the part. So you can print as many wheels as the bed will fit in the same time as one, which should only be about ten minutes on a resin printer. Resin also has a lot more flexibility in its properties than fiber printing can, the toughest printing resins are tougher than any fiber for this application, so the parts would last longer.

Once you have the process dialed in, you can probably print these in layers, which can be clipped apart and UV treated in batches, or maybe just use a magnetic fixture if you don't mind attending the machine more closely, they'll just pop off of one of those.

It reads like #8 sent you off on a substantial mission because of printer fidelity in the registration spokes, a resin printer would let you explore that design more thoroughly.

I think an alternative approach would be to have some sort of woven fabric loop that can have raised braille dots stitched into it from the right side and then the loop is pulled to the left. So it would read like one of those scrolling LED banners except it would be a fabric belt with raised dots.

Maybe the simplest approach mechanically would be to have Nitinol wires running vertically, parallel to the 'weft' thread. To print a single letter in Braille would require stamping two Nitinol wires to render the raised dots in the correct positions, and then feeding the fabric belt to the left. On the return side, hidden underneath the display, the Nitinol wires could be returned to their original state.

Alternatively, any technique that could raise a knot in the fabric which could then be easily released would be worth pursuing. That could be as simple as something that pushes a loop up (for a braille dot) through the fabric belt and then pulls it back out on the return trip.

Unless I'm missing something, this article seems to assume 6-dot Braille cells. But every Braille display I've ever worked with uses 8-dot cells; that standard is called computer Braille.

Dumb question. Could one instead go with some sort of inflatable system? Thin layer of plastic with some controllable channels. Inflate the individual dots? There was an early smartphone that had a system like this for tactile feedback, I don’t know if it ever went into production.

Very good question, just came across - https://www.pnas.org/doi/full/10.1073/pnas.2106553118 which seems designed for braille.

Very cool. I love seeing things like this being made. Some others from the past few years:

https://techcrunch.com/2017/05/05/blindpads-tablet-makes-vis...

https://techcrunch.com/2018/01/18/the-becdot-is-a-toy-that-h...

https://techcrunch.com/2022/03/10/dot-pad-tactile-display-ma...

https://techcrunch.com/2023/03/17/the-monarch-could-be-the-n...

Something I've heard is that the content, APIs, etc are equally important, or you end up with a great device with nothing to display.

Assuming actuators are expensive. The best approach must be to reduce the number of actuators to 1 main driver to cycle through columns plus 1 toggling actuator per row. (Further reduction into a complete XY-board will be too slow and complex). Move the memory to the mechanical domain.

That will reduce the number of moving parts to 1-2 per pixel, while still keeping the actuators complexity rather simple. The ballpoint pen isn't a bad idea, it just requires refinement. With some clever engineering i bet you could make that out of one single plastic part that can act both as a spring and a locking "click"-mechanism.

Similar to the Nist wheel driven display linked in the article [0] but invert it, such that the wheel rolls over a flat grid of miniaturized "ballpoint pens" to set them in the desired state. Rendering a full screen will take 1-2 seconds of rolling the wheel over all the columns. This should be an acceptable compromise.

[0] https://www.youtube.com/watch?v=sBBqdEnZPck

Fantastic effort, but oddly and almost shockingly archaic. This is seriously the way the problem is being solved? Why isn’t it microelectronics or MEMS? If we can print 3μm OLED pixels surely there’s a way to lift a discrete 1.6mm dot, especially with that huge 2.5mm spacing. Even adjacent dots ought to be no big deal. Maybe a haptic kludge or grid of air pressure nozzles?

I have a personal theory of invention that states: “If you thought of it, someone else thought of it.” I’ll go have a look…

OK, an hour later, what about this Dot Pad? “Dot actuator” is micro mechanical? Looks like almost a product. Dig the groovy scroll-in-place presentation, wtf.

https://www.dotincorp.com/page/31

Only £12k! https://www.visionaid.co.uk/dot-pad

Seems to be £12k, so not cheap: https://www.visionaid.co.uk/dot-pad

I was thinking if the dots need individual actuators. I'm old enough to remember dot-matrix printers; take the print head but put slightly beefier pins on it, move it left and right and use it to push metal spheres up in a hour-glass shaped cavity, with springy walls in the center section.

Obvious problems: how do you reset the spheres? And wear issues.

Edit: You could replace the hourglass throat with a plug valve (shared between pins). Print head comes along, turns the valve open, pushes the necessary balls up, closes valve, moves to next column. Needs more mechanics per cell, though.

Your desire sounds a bit like the fluid based, piezoelectronically pumped haptic display presented by Shultz, et al, in 2023 [0]. I can't speak as to whether these could be packed in enough. The video indicates they can make cells of 2mm wide.

[0] https://www.youtube.com/watch?v=j_rErbhxNFM (4 minutes

A more usable page: https://www.dotincorp.com/page/32?gbn2=DotCell

Doesn't look like it's MEMS? Unless each individual cell is a MEMS, but that seems strange. Maybe MEMS devices can't generate sufficient force?

"MEMS actuator braille" has many research hits, but no obvious product.

They have a bunch of patents explaining the tech https://www.freepatentsonline.com/20240054912.pdf

The device people say cost 12k does a lot more than just braille so that's not a fair comparison.

It seems something like the media produced from am embossed labeler or stylus could work. Dot matrix printer head could be a good starting point. If the medium was vinyl like it possibly could be reused by feeding it through warm rollers to the print head.

please consider this reference https://inhabitat.com/ted-moallems-braille-it-labeler-is-mad...

Did that come to anything?

yes, it was fully vetted and open design that had a strong acceptance for the vision impaired. The following Design criteria were considered: 1. durability 2. reuse, repairs and maintenance 3. manufacturability 4. adoption and adherence 5. skill level of the target population to use the device 6. price

https://hackaday.com/2021/06/27/mechanical-7-segment-display...

This is the approach I would take to this problem. It could be miniaturized a great deal.

Notched cams seem like an obvious win here.

Ctrl+F "dot matrix". Not found.

How do printers manage to stick dots into paper? Is it similar to any of the ideas in the article?

I wonder to what extent dot size / spacing matters. Traditional braille is absolutely tiny to make it even remotely possible to construct books, but with a digital device it might make more sense to make it larger.

Making it 3-4x as large would make it fairly trivial to use off-the-shelf servos for each dot. Instead of moving your finger left-to-right across multiple characters, it'd raise & lower the dots while your finger stays on them.

You could even take it to the extreme and create a "reverse steno" display: imagine a regular keyboard home row, with one dot per finger. Use the thumbs for speed control or word skip or something. Maybe it could do double-duty as input method.

What about "printing" braille characters on a continuous rotating reusable tape?

Those silicone "popper fidget" toys come to mind as something that are flexible, could be cheaply mass produced, and could hold a given state at the "print" head. It might be possible to embed small metal ball bearings inside that are moved between the outer & inner surface to increase tactility.

The print head just needs enough actuators to change the state of a given column of dots as the tape is rotating past, and a feed motor to move the tape.

A portable and affordable braille reader would be great. I had a couple of blind students in my calculus course a couple years ago, and screen readers are just not a good substitute (too slow and hard to understand when there are a mix of words and equations). This is especially the case in timed assessments like exams and quizzes, but even for homework it is a serious obstacle for the students.

Now, if there were only a FOSS tool to compile tex into Nemeth...

Considering the human finger is very sensitive to texture perhaps there is a material that change its texture in response to voltage?

Why not just use mechanically decoupled pins that vibrate with a smartphone vibration motor? You don't need texture if the thing vibrates.

Cost and noise would make that extremely unpractical.

As opposed to the low cost texture changing material proposed?

If I had to do this I'd probably look into PCB-based solenoids with compliant mechanisms. But there are already people working on that.

why couldn’t it be done on an off-the-shelf 28-byj-48 stepper?

The reason is probably explained somewhere in the article, I just can’t find where

The 28-byj-48 were designed to oscillate air conditioner vents, and as such have an internal slip-clutch mechanism to add to the abysmal backlash of their gear box.

There is also the weight and energy consumption to consider.

People may be better off gluing a small bead to one side of a flip dot display. =)

I took a look at creating an open source solution in this area last year and in fact someone sent me this page last week. The biggest challenge is spatial. The braille standard places pins very close together.

In modern commercial braille terminals, generally a large assembly is offset from the braille block in order to support it. However, this is not feasible if you want to have a dense array of braille blocks. The primary reason for the bulky designs is that they rely upon piezo-electric crystals which move only a small amount and thus require long levers to actuate the required length. Such a mechanical configuration limits you to two lines of text close together at a maximum.

Another category of solution explored is wheel based solutions, a well established category which fail predominantly on density. Specifically, if you cannot place two lines of text very close together vertically then you are going to have problems providing a significant amount of information at once, because you are either going to have 1-2 very long lines or lines will be so far apart as to create an enormous matrix with reach issues. They are also sub-par on mechanical complexity, aggregate weight, refresh speed, and assembly cost due to part count.

Two alternatives are pneumatic actuation and electromagnetic actuation. The problem with both is that there is no standard solution suitable for small size / high spatial density required, but perhaps one may be developed. https://youtu.be/k1inMrAZ_Eo?t=45 is an example of an actuator created within a relatively small size which would be potentially inexpensive to deploy. Ultimately, data comes electronically so pneumatics may be seen as an expensive and complicated middle-ground offering lower weight as its primary benefit but probably sub-ideal as a primary focus of research due to additional cost and complexity.

My analysis therefore concluded that the future of braille displays lies in on-PCB coils for micro-actuation to reduce cost, weight, and part count while increasing density. This approach was inspired by Carl Bugeja's videos. Since August 2023 I have a working prototype of sorts using commercially available dowel pins and multiple PCBs stacked in a vertical structure. The initial prototypes of this onboard electromagnetic coil based actuation system showed great promise but remain to be validated. Chief concerns are torque and a locking solution, the latter being important for reducing power consumption, though fallback strategies exist (eg. hand tracking or 'presence detection'). I am currently focused on other areas but would like to return to this in due course. I would be happy to open source my work (BOM, KiCAD, OpenSCAD, notes) if others are interested to take it forward.

The current prototype is as follows.

                                 ∩  ∩     ∩  ∩
   ==∩==∩==∩==∩==∩==∩== PCB 1  ==‖==‖==∩==‖==‖==∩==
     ‖  ‖  ‖  ‖  ‖  ‖            ‖  M  ‖  M  ‖  ‖
     ‖  M  ‖  M  ‖  M            ‖  ~  ‖  ~  ‖  M
   ==‖==C==‖==C==‖==C== PCB 2  ==‖==C==‖==C==‖==C==
     ‖     ‖     ‖               M     ‖     M
     M     M     M               ~     M     ~
   ==C=====C=====C===== PCB 3  ==C=====C=====C=====

      Position at rest.          Ejected position.

PCB 1 functions as a cheap precision guide for the pins.

PCBs 2+3 function as coil bases for the pin actuation. Two are required because the required coil density cannot be achieved with only one PCB. (This necessarily means half the pins are longer and half the pins are shorter.)

M is a small circular magnet, glued to the base of the pin.

C is an on-pcb coil at close to maximum density. When C is energised, M ejects with great force, but because M's diameter is greater than that of the pin's PCB hole, the pin-magnet subassembly stops its vertical travel at the predetermined position.

~ is an active coil field.

Benefits of the design: cheap, uses commercially available components, low weight, achieves required density.

Challenges remaining: Energy consumption, simple cheap control plane to maintain density, locking mechanism (non-energised mechanism for maintaining raised state), manufacturing process development (automation required for cost-effective output due to large number of pins).

References: http://libgen.rs/scimag/?q=braille+display