Neltume: Language learning device for blind people

After testing and carrying the device for some days with me, I realized that the size of the second prototype was too much and not really comfortable, so I decided to test not only decreasing the size of the case itself, but also changing the buttons.

And this is the comparison with the second prototype.

I tried yesterday to print the second part for the case, but there were some problems with the filament during the night and today the print work wasn't totally finished. It was good enough to use some tape to close it and now I can made some test with the different sensations and dimensions. I am planning to cut with a CNC the next prototype.

I moved back from the country were I was studying abroad and developing Neltume during the last year (Germany) to my Homeland (Chile) and during the process I had to turn apart the prototype that I initially made out of a Christmas Box.

Last Week I decided that I wanted something more elaborated and in the FabLab of my University I was allowed to 3D print a new Prototype. It was much much slower than my expectations and totally unpractical for futures test prototypes, so I plan during the week to use a CNC Laser Cutter in order to create a new enclosure.

Also I tested the C.H.I.P. Pro and after some changes in the the button controller script (that was initially thought for the Raspberry Pi) I could make it work perfectly.

To make work the GPIO ports of the C.H.I.P. Pro I used the library of xtacocorex and it works perfectly.
Some friends of the Fab Lab gave me good feedback about the interface and usability, like the size and position of the buttons. I am reconsidering the form factor, it could be a good idea to use a Joystick to control the navigation and selection option and it would be interesting to design a prototype with the form a usability of a NunChuck (Nintendo) so the user can control the whole system with only a hand.

Other possibility is to replace the actual arcade buttons with small micro push switch. I like how it looks the design that the people of Adafruit created for the BMO 3D printed project and it could decrease the size significantly.

 

After trying unsuccessfully to improve the audio output of the Raspberry and seeing that I was adding more and more components to the project, I found a very good alternative to the Raspberry Pi, CHIP Pro.

With CHIP Pro I don't need to  buy an external LiPo Charger (like the Powerboost 500) and an external USB audio output, because it has a digital audio output. This combined with a lower price, makes the CHIP Pro an excellent alternative to the Raspberry Pi.

Now I am doing the migration to the new micro processor and a new Enclosure for the prototype.

Navigation:

• In the moment that you access to a menu/submenu it is said the name of that menu and goes back to the first option of that menu, exception in the submenus where you go directly to the second option (the first one lets the user go back to the previous menu)
• When you press button 4 or 5 (top buttons) you can directly go back to the previous menu (it was a pain in the ass to go to the first option to go back)
• Add the option to review the button enumeration

Reviewing system (there is a lot of work to be done):

• Before it was used a very basic system, where the worst word was always reviewed first, now I am implementing the Leitner System for spaced repetition, which now contains 4 reviewing options:
  • 1: Review 2 times the day
  • 2: Review 1 time the day
  • 3: Review 1 time every 2 days
  • 4: review 1 time in the week
• When the user turn the device on, it is checked if the day is the same as the last reviewed word and check which words need to be update (change the leitner box

After reading documents about the situation of blind people on Latinamerica and specifically in Chile, where only between 15-20% of the total blind population can actually read in braille or other systems, a rethink and redesign was necessary at least for an early Prototype.

After rethinking the actual design and button distribution of the device I changed the distribution to 3 buttons (bottom) for navigation (Right - Confirmation - Left) and 4 buttons (up) for multiple alternative selection.

Prototype B is implemented in a Raspberry Pi 3B, because I can easily transfer the file through SSH via WiFi.

The sound of the Raspberry Pi via Jack 3.5mm is just horrible and noisy, one alternative is to use a USB Audio Card.

A lot of variables aren't really considered for the first prototype of the learning and reviewing system.

One of them (and very important) is time and the generation of reviewing frequency.

Until the first prototype and the arrival of the RTC there is no option for the tracking of time on the raspberry pi.

Also, there is a difference between the actual learning system of the alphabet. I was dealing with the amount of letters to learn per session, and the exercises were a randomly distribution of those letters. After coding the word learning algorithm, I realize that a big exception (when you were in letter 24 you only have 2 letters left instead of 6 for the rest of the code) could be skip.

So, the person learn one letter, and the other five (to complete the six buttons) are randomly selected from a the alphabet (which doesn't include the letter that the person is learning)

Also, there is still some general problems with the conception of the reviewing code. One of them, is that I'm actually not tracking the number of right/wrong answers. Should it be in the same json file or another?

Example of how is going the development of the audio interface and word learning process. Everything is still in my computer, waiting for the arrival of the buttons,

I have already generated some of the basic functions available in Prototype A Software

• Learn Alphabet
• Review Alphabet
• Learn new words

In order to keep logs (with dates) of how many times the user had used Neltume it is necessary to use a Real Time Clock (RTC) that raspberry doesn't include in his original design. So, until the RTC arrives we are only using this format make a log of the word:

log –> #id_word:[#goods, #total]

In this first prototype (A) I want to develop a laptop only interface. Using the keyboard distribution previously shown on Log2. Prototype B will consist of a model inspired on the braille-keyboard, with 9 main buttons a.k.a. Neltume XL and Prototype C will have different configuration of buttons, and possibly some of them will have an arrow shape to navigate through the menu. Maybe it could be an easier interface use only four option button (instead of the six actual), two for navigation and one for confimation. Only tests can confirm which one is easier.

Also, thinking of other possible uses for Neltume, it came to mind the possibility to use it like an interactive audio book and for test preparations.