 Hi, everyone. Today, me, Nelit, and Varjan will be presenting our project for Capstone Braille tablet. Short outline of our presentation, there will be a project overview, design methods, final results with a successful demo, hopefully, and future improvements of the existing product. So project overview. Just so we're all on the same page, what is Braille? Braille is a tactile system created to assist blind and visually impaired people to read and write. And so you can see here are the letter formats. And what was our goal for this project is create a prototype tablet, which will use these characters as letters to make kind of a kindle for blind people. We wanted it to be low-power consumption, portable, and, most importantly, affordable. And now how we achieve this, we'll discuss in design methods. We have free sections, mechanical design, electrical design, and software design, so let's hope into it. So the very initial design of the pen was the following. This would be a non-metallic core, a magnet here, another magnet here, and two metallic washers here. This magnet would be stationary held by this printed cover. And the pen would move inside of it. How would that happen more specifically? So there's an inductive coil here. So what would we do? We would supply power to the inductive coil, so it would have a certain direction. And because there's a magnet here, it would, for example, push up the pen. And because this is a magnet and this is a metal, they would be attached to each other. So after this movement happens, we would no longer need power supply for the coil. The pen will retain its position. And the same, we would just change the polarity of the supplied power, and the pen would go downwards and mix the directions, but you get the idea. So this was our idea, how do we low power consumption? We just need to be able to move the pen. And after that, no voltage is needed for it to operate. So we had several experiments. This is our very first model. Not too good, we think, and the setup of the experiment wasn't that well. We used a printed filament as our core. We drilled a 1 millimeter hole in it to put the magnet inside, as was seen in the diagram. And this was our printed core. And we wired the coil by hand, very badly, very non-uniform. With supplied 5 volt powers, it consumed 500 milliampere current. And we used 0.09 millimeter wires and 0.5 millimeter wires. None of the cases, we were able to actually do the required movement. We need to detach the magnet from one disk and detach it to the other. Just the coil would move because of the magnetic force. So we realized that we need to, of course, improve the core. So the first thing we did was replace the plastic core with the copper wire, because then we would supply power to it. It would heat up and deform, which was not ideal. And then the cover was also changed so the pin can actually stand up to make the experiment, the environment of the experiments better so we could understand what our actual problems are. And also a place for the coil. So we could position it with respect to the magnet. So what we did also, and still when we supplied it with some power, it was not enough to pull or push the pins. So what we did just for experimenting, we put a plastic piece in between the metal and the magnet to kind of decrease the force of attachment between those two. And we gave it, we used a 0.22 millimeter wire, which has high, which has slower resistance because of its thickness. And gave it 12 all supplying which case the coil would consume 0.8 ampere of current. And finally some semi-successful test. This was the first time we were able to actually move the pin here and here. So, but again, this was not ideal as we didn't want anything to be in between the magnet, the plastic. And we realized that we needed a stronger coil. So, sorry, yeah. So what we did was actually add an iron core to the inductive coil to make the magnetic force stronger. And because there's a magnet at the bottom of the pin, the pulling, the ability, we were able to pull it much more easily. However, because there was a nail inside that the magnet was attracted to, it was a bit harder to detach the magnet from the top metallic pin. So what we actually did, we removed the metallic pin and placed the plastic one instead as it was still because of the attraction of the bottom magnet and the nail. It was still able to retain its position even without any power supply. So again, 0.22 millimeter copper wire was used totals of eight meter and we were finally able to achieve reliable pin movement. So this is a more detailed dimensions of our final product. This is a cover of three sections designed to hold six pins together which would make an entire character. And this is the actual look of a character. It's a bit bulky. This is a size of standard pin for braille which are used on paper for reading. And this is the size that we got which is, you can see significantly larger. You are still able to feel all the six pins by a finger but still big difference. And we will discuss how this can be improved in the later sections to move on to electrical design. Now onto the electrical design. Initially our goal was to create a 14 character tablet. Why 14 characters specifically? Because 99% of all English words are comprised of 14 letters or below. And hence why we decided to do 14 but at the end of the day we were only managed to do one just because of time constraints and so on. So, but the problem with 14 also characters is that if we're doing 14 characters each character is gonna have six pins that we're supposed to control. And that will be a total of 168 output pins. To counteract this, we use shift registers instead. So basically what a shift register is it's basically a electrical component where you just serialize the data transfer and to make it more understandable in simple terms basically whenever you have a single input you can create multiple outputs via multiple clock cycles. So you just insert a serial data and then it outputs multiple outputs multiple currents. In this case we also use H bridges. So the H bridges was used to control the direction of the pins either pulling them in or pushing them out. Here's the semi-final circuit design. It's not the final design because there we also added a SD card SD card adapter to the Arduino but you can see that there is two buttons over here which will be the user input. One of them will change the page to the next word in this case and the other one will do the opposite go to the previous page. We have two shift registers and as you can see one of the shift registers we call it the inverse shift register. Why is that is, well first of all whenever we're putting in data if we have a data of one bit into the shift register then the inverse will give a zero data and with this combination of one and zero if we implement it into the H bridge we can control the direction. So in this case a one and a zero would go push it upwards which we identified as the red LED. The LEDs kind of are a stand-in for the coils and of course vice versa if you put zero and a one then it will pull it down. This is the shift register serial data transfer diagram so basically this is how most of the logic works. So we have our Braille code in this case a one would mean that it is being activated so it's either pushing upwards or pulling downwards and zero means that it just doesn't do anything. And then we also input a into the shift register a high bit value so in this case a one. If the first pin, the data of the bit of the first pin is equal to the data pin input so in this case both of them are one we will enable the shift register hence the current will start flowing. Then after the first initial we insert a zero next to the bit and you see that the value of one kind of shifts to the next pin and then we compare that to our Braille code in this case it's different hence why we don't have a enable input and the amount that we keep the enable input depends on how much it's necessary for the actual physical pin to be pushed. And with our experiments what we achieved was actually 10 millisecond impulse was enough to push the pins up or pull them down. So we did approximate calculations of what a battery operation lifetime would be, what a battery operation time would be in this case it's not with like seconds or hours but how many pages can be displayed. We took 14 characters as our page number so current consumed when we're pushing or pulling the pin with this design is 0.9 amperes. Time it takes to display a page of 14 characters would be 10 milliseconds times 14 times six and we supplied with 12 volts of voltage. So energy consumed is a number there 9.1 and we took a battery that would satisfy our needs which is 14.8 volts and 5,000 milli amperes per hour and overall energy in that battery was the following number and then we divide the energy in the battery by the energy in the switch. We find that we can display around 29,000 pages of which is 29,000 words of length of 14 which is on average like one-third of the book but will be improving. So not onto the software design. The best way I can explain the flow in this we can divide it into three stages the pre-initialization stage, initialization stage and the display stage. So during the pre-initialization stage how are we going to access the book? So we have a micro SD card that is inserted into the tablet itself and this is where a user helper would be very much useful because a blind person cannot figure out to work with a screen. So in this case someone who can help the blind person can insert their book of choice and it has to be named book.txt at least for now. They can just replace the inside of the text file with whatever text of book they want and also for future implements we're also gonna have a config file. Why is that? Just because if we have different tablets with different character numbers then it can change the way it works to optimize. Here is the initialization stage. So whenever you start the tablet it goes through a process of taking out the data from the book, the text and then transiting it to Braille. Hence this is the part of the covariate transits and if you look over there there is a little snippet on how the mapping works. So we have a binary data that will represent our Braille code and next to them will be the letters that we're accessing. So the number of whichever row we take that is the corresponding letter. After the initialization stage comes the display stage which is divided into two sections. The first is a converting of electric electric to physical data. This is the logic that I explained earlier but only in code form. But before it gets to that it first checks the position where the user is into the book because we don't want to lose where we are whenever we're gonna go to the next page it has to remember. So in this case it looks through the data and it checks where the position is in the text file and then based off of that it takes the number of characters that we have on the table. So in this case it only takes one character takes that the translated character and then goes through the iteration. And then comes the user interface which basically it's a very simple code of whenever someone whenever the user presses a button it changes the position and changes the position so that we can take the next iteration so in this case if I press next it will change the position by the number of characters it will take those translated data back and continue henceforth. So our final results what we achieved so far that we are able to control and move each pin using only a 10 millisecond of impulse and after that the pin is able to keep its position without any power being supplied to it. Currently the entire system can be supplied by this 12 volt 5 ampere power supply but a battery but the same characteristics can be used to replace this so a socket would not be required. Just the better requirements that we found out where we found it quite late in the stages so we weren't able to order those batteries. Battery operation time was 29,000 as we said the dimensions of the character a bit bulky as I said 28 by 14 by 26 millimeters base price per character. So base price considering the controller the controller the SD card the wires that are also used to make the coils is 86 and 2.6 dollars would be per each character. So if you want to add character you only need to add 2.6 dollars and the system is modular because of the shift registers which when they get the data you can just take the wire and put it into the next shift register. Also the H bridges that control the spin are independent of the H bridges that are of the other pin it can just like a puzzle piece just come and stick into it and with the code and the config file just change of the amount of characters and based on that the code would then itself divide the text into the appropriate amount of characters. So let me try to do a display now there now, okay. So upload the code the bond is connected. So here we have the switch and I don't think you can actually see it but here this one turn which one came up and then this one and then this one and then if we go back it would be the previous character. It's a bit hard to tell but you can definitely feel the difference it's very minute but it can be felt and some because they are so low improvement from our previous designs is because they are so low then you just touch them it's not very likely that you push them down unless you physically try to do that which is definitely a plus. So this was our demo unfortunately I can't move it up there but if you would like to come and test it afterwards you are very welcome and so our future work and improvement so the main mechanism here is the pin movement itself. So every component that we chose was based on the pin movement and the requirements that arise from it. So what would make the pin better? If it was symmetrical before I say this I should mention that almost every part beside the magnets was made by us. The wire was cut by us we would file it down then we would attach it to the magnet we would make the bottom metallic disk on which the magnet is to be attached the coils I was also wired by ourselves so there are some human imperfections so the bottom magnet should also be exactly lined with the pin which is kind of hard to do then you're working with such small components if the pin is smoother there would be less friction between the pin and the magnet for which it's moving and all the pins should be identical so we don't have to do some tweaking for each pin specifically. So why would this help us? If they are symmetrical and straight less force would be required to push or pull them because sometimes because of the difference in weight or asymmetry instead of going down it would just go sideways because at this part some part is heavier than the other so it wouldn't go exactly down so we can minimize that. If the force is less then the coincides can also become smaller because we won't need that much current and so making the tablet smaller in size and lighter and less current consumption as I said and smaller signal pulse would be required so we would have a better battery life so more pages could be displayed. This is it, thank you. Instead of the shift register I'm not trying to redesign it, it's just a question would you just use a software control parallel approach? Software control parallel approach. On the process or I suspect it must have that. You probably could, I'm not sure but probably. For now, that's hardware, that is true. That's hardware, the fact that matter is for us we mostly just wanted to focus on the mechanical parts of the project and I do not have the coding expertise or. Oh, it's really easy. But the idea of shift register was simple enough that I know how to implement it and we just work with that. And the second question, has anyone else ever thought of this? This design specifically. No, I mean does anyone, is like a braille tablet available by the Chinese or something? There is a braille tablet, it was actually mentioned in our connoisseurs, one that I used but I could not find the specific mechanism that they used. That was the only commercial product that I could find. Which was in some aspects better than ours that the pin sizes were smaller so they could display more characters but also it was much more expensive, about $3,000 and also it had, yes. Yes. And also it had to have. We are having big numbers by a lot. Yes, and also it had to have constant power supply which in this case we do have a power supply but this can be replaced with a battery which is not the case for their design. So pluses and minuses. Another tablet design that uses instead of electromagnets it uses piezoelectric components actuators but each one of those pins costs a dollar. In our case we could have done a whole character with $2.6. Imagine we had six pins that would result into $6 and that's almost like doubling the price in that sense we're cutting the price by a lot. So the piezoelectric is buzzing on your hand? Is that what it's doing for it? No, it's pushing it up or pulling it down. It pushes it back up, is that what it's doing? No, no, no. We can't forget it. Go ahead, next. All right, all right. I haven't got a question. Do you think, would you invest into this kind of a project as a business? Do you think in the long run any time braille tablets are going to be actually applicable? But do you think it's both of you? We hope so. I kind of thought about it, kind of thought about it, we're still into the... For now we were just, okay, let's just finish the capstone, bad kind of mentality. But we have a lot of time after this. We're sure to figure it out. It happened because of the other technologies being developed, like if the person isn't like both blind and deaf, I mean otherwise you can actually hear everything. That's how everything develops. I actually mentioned that in our connoisseurs that there are differences between when you listen or when you read even just by touching it. There are differences in how you perceive. How you can perceive the, I don't remember the exact, so we're doing a lot of brain activity stuff which I'm not an expert of so I don't remember exactly but the way the information is retained when you read it instead of listening to it is a bit different. Another thing to mention, there are audio books that exist but people still read. There is a way, there's also preference involved in a lot of stuff because you might read a book in a different intonation in your head than whenever you hear a person talk instead. There is a lot going on. So in this case some people just prefer them themselves just hearing their own story, their own voices and making their own characters in that sense while they're reading. So that kind of also helps. So this was about giving them the choice, so making it affordable so people can have the choice between having the audio and having the actual physical thing because both are great but it's like better than you can choose between two. Are you aware of the market, like what's the market size? How many, what's the size of? Very, very small that we could find. That actually worked because they found a lot of projects. No, no, no, no. Oh, the number of people who need this. The central market, potential market size. We didn't do that much research yet. Again, we were mostly on, for this part, we were just mostly focused on the engineering part. Later on, then this is where we're developed, we would of course have to do that research as well. My question is about the engineering. You spend a lot of time engineering the character. Yes. Have you heard about the concept of not a matrix printer, which is moving piece? Okay, with a very, very small device that you could just simply pick a place and do your kind of character generation. It's very small, but it will easily miniaturize your device. Oh, can you say the name again? Dot matrix printer. Okay, thank you. It's the oldest type of printer that you can. Yeah, okay. It's what we used to use. Okay. It's very old age. Back to the classics. Yes, I have two questions, but before I get into my questions, I'd like to add some points to the questions that our colleagues here asked. There are two US-based companies that actually became developing for the tablets for the past eight years, but either one haven't reached the commercialization stage. One of them is still in the financing ground, and it's not obvious if they do a quick online research. It's not clear what are the obstacles they are facing or they've faced before that over the past seven, eight years that they haven't been able to commercialize this product. So I'm also very curious to get into that. As for why a great tablet instead of an audio book, mostly, I'm not an expert, I'm not a neuroscientist, but it's coming from the biomedical engineering domain. Whenever there's an impairment in the brain, other areas in the brain also get affected. So whenever someone has some hearing difficulties or blind, those closely correlated regions of the brain they get affected as well. So it also depends on the preference of the patient, of the end user. One of my questions is that in one of the diagrams, you know that the distance between those pins was 2.7 millimeters. Yeah, this one. Yes. Is this the closest distance before the magnets into different pins start affecting each other? No, this is a standard print on, standard that is printed on a book, a braille print, that's what we're comparing to. The magnets don't really affect each other because, Okay, let me ask this in a different way. What's the minimum distance between two pins before a magnet in one pin starts affecting Okay, so what about the code? Whenever I go back, can you go back? Just physically for the magnets attracting each other, at the moment, we couldn't test them very closely because the size of our coils is 7.9, 7.8 diameters. So imagine that plus one millimeter distance is the most, maybe a bit less than one millimeter distance. So it's about eight or nine millimeter distance that we can actually test. Otherwise, the coils would touch each other. So we don't actually yet know how close the actual magnets can be to each other for them not to affect. But you're in the ballpark, you have an idea. Yeah. Also, there is another thing to kind of balance things out. So whenever you look at this code, I didn't mention it, but we're actually not doing a serial input parallel output in that sense. It's all serial input, serial output. So in that case, all the pins are going one by one. So why we did that is just so that whenever one pin, if two pins are gonna go up at the same time, the magnets would not kind of ruin each other. In that case, might not affect, the magnet field wouldn't affect each other. So in this case, that's why we're doing one by one to just minimize those little errors because it might push it to the side and the pin might do the friction, might not be able to pull up. So in this case, that's why we did serial input. Yeah, about that also, not the magnets of two different pins, but in this case, for our design, because it's very, it's like this size, this magnet and this magnet were really close to each other. So we actually did something to help us with the movement. We put them in the same direction. So these don't attract each other. So our movement in both directions is easier. So that's the only experiment with the magnet interaction that we have done so far. My second question is that it's more of a general term. If you have more time to work on this, what would be the first point you would try to improve or try to focus on? I don't have a different answer than me, so you go first and I'll go first. Yeah, probably the design of the pin is very important, as I mentioned. We don't know yet what kind of, it needs to be automated in some way. We know that definitely because, first of all, it takes incredible amount of time to make just one pin, let alone a lot of pins for a lot of characters. But also as I said, they're not really symmetrical, so they have a lot of imperfections. So I think what we would focus on is research how we can actually make those. That would be the first thing. So, Christ, you mentioned it's just the bomb price, okay? It's not the material price. Yeah, not the work. Material, material. Yes, it's not the work. It's not the work. It's getting into the business. Oh my God. You are spoiling the engineers. If we get access to a lot of industrial-made products, it will definitely make a lot of things easier because the magnets might be, again, might be a bit strong. We might need much weaker magnets. But of course, there is not a lot of production out there for this size of magnets. And then now we're trying to be picky of how strong the magnet is. There is a lot of finicky things like that. If we could, we would change them. And as well, the same thing with the coils. We tested it after we finished this tablet. We realized that we could even go one, make the wires a bit thinner. And so that will also kind of change the size and might bring the pins even closer. Yeah, so one of the pins has a coil diameter of seven, which is the smallest one that we got so far. But we got that for like two days ago because one of the wires got loose and we didn't want to waste time to make five more again because we needed to assemble all of this. This is a device for future. You could make uneven length of pins. So you could put them a lot closer. Yeah, yeah, we were thinking of that. Just the coil. But the research taught Patrick pretty well. Yes, yes. We'll see you in a second. I'll go home first thing. That's all I want to do is research. I have to interrupt. All right, see you. Thank you. Thank you.