John, I have copied you on this incase anyone wants to follow up on your message below. In light of what issues are currently under scrutiny this is good information Thanks! ----- Original Message ----- From: "John Roberts" <john.roberts@nist.gov> To: "David Poehlman" <poehlman1@comcast.net> Sent: Tuesday, March 12, 2002 9:49 AM Subject: Re: smooth scrolling... braille output hardware (jf) (LONG) Hi David, I'm not a subscriber to the blinux list - please feel free to forward this to the list and/or to interested individuals. I believe that the smooth scrolling as described by Jan Finegan has excellent potential for accessibility technology. We at NIST have been looking at it as one implementation of tactile graphic technology (and filed an invention disclosure on it with the Patent Office a month or two ago, so I can finally talk about it). If the spacing of the stimulus points is sufficiently small and the response time (refresh speed) of the array is sufficiently fast, it can be useful for both graphics and Braille text. A lot of research on this has been done at Johns Hopkins University. They built an array to stimulate one fingertip. Their objective was to study many aspects of the sense of touch - the device is much too large and expensive to be a commercial accessibility product. However, I believe they were able to accurately produce moving Braille text that Braille users could read. They used a 20 by 20 array, with spacing between the stimulus points of 0.5 millimeters, so the total array fits 400 stimulus points into an area of about one square centimeter. Since the spacing between Braille dots within a standard Braille cell is 0.092 inches, their display does about five scrolling steps across a single Braille character. My opinion is that this spacing would probably be a good starting point for someone trying to build a smooth scrolling display - if anyone can find funding for research, it would may be useful to try to find out whether a larger spacing is acceptable, and whether a smaller spacing produces better results. The Johns Hopkins project page (which was working recently, but I don't know whether it's online right now) is http://www.med.jhu/somlab/lab/research/methods/400pin/ As you know, for the detailed sense of touch to work, there has to be a sensation of lateral motion between the object being sensed and the skin. In a conventional Braille display, all of this motion is produced by the user, by moving the fingers over the Braille text. In the NIST rotating wheel Braille display, the majority of the motion is produced by the motion of the rotating wheel. In both of these cases, the Braille dots are physically present, so the spacing of the dots conforms to standard Braille dimensions. In a smooth scanning display, I believe the best approach is to have no real lateral motion between the fingertip and the tactile stimulus array - the sequential stimulus of adjacent points creates the illusion of lateral motion of the fingertip across a surface. Of the four types of "mechanoreceptors" in the fingertips (which among other names are called RA I, RA II (Roman numeral 2), SA I, and SA II), the RA I receptors are most important to a detailed sense of touch, and this kind of stimulus can do a very good job of creating the sensation of a detailed tactile surface (which can include Braille text) moving under the fingertip. The issue of how the user controls the apparent motion of the displayed material is very important. The sense of touch is a scanning sense, meaning that people are used to running their fingers over a surface, and mentally integrating the sensation their fingers receive along with the position of the fingers, to create a mental model of a 2- or 3-dimensional object. One of the most intuitive guidance methods is made by mounting the tactile display on a tracking device such as a mouse. As the user moves the mouse, a computer tracks the position of the mouse and updates the display to match the virtual object at the corresponding location. This method preserves the connection between finger motion and the sense of touch, and allows the user to move freely in all directions. This method has been used by VirTouch Corporation in an accessibility product that is commercially available. Their web page is http://www.virtouch.com Two related patents (6,278,441, and 5,912,660) describe some of the ideas they had in designing the tracking mechanism of their mouse - for example, I believe it senses rotation of the mouse on the surface, as well as the up-down and sideways motion that conventional computer mice measure. Their product has three fingertip displays on top, which I believe are 4 by 8 arrays of stimulus points. The spacing between the stimulus points is much larger than the Johns Hopkins design, and I don't know whether they are effective for reading standard dimension Braille. At NIST, we are interested in a high-resolution fingertip display suitable for both tactile graphics and Braille, that can be made small enough to mount in a mouse-like tracking device. We are also interested in one that can be mounted in the fingertip of a virtual reality data glove - the detailed sense of touch in one or more fingertips, coupled with existing force feedback technology, can add an important new capability to virtual reality systems, both for mainstream applications and for accessibility. I gave a talk on this at the SPIE Photonics West / Electronic Imaging 2002 conference in January. Hopefully within the next few months, we will be able to put a link to the paper on our Braille web page. The paper discusses these technical issues in more detail, including what we think are some of the main challenges for implementation. Jan made a reference to needing reading space for at least two fingers. Some of the Braille users I know use one finger to read Braille, some use two, and there are many other combinations. It's very difficult to build a display that would match everybody's favorite reading method. The Virtouch device has displays for three fingers, but there are spaces between the fingers, and some users say they would rather have the fingers right next to each other. If the driving mechanisms of the displays can be made compact enough so that they can be mounted in the fingers of a data glove, then the user can quickly change the spacing between the displays just by moving the fingers apart or closer together. It's important in any design to maintain continued communication with a fairly large group of potential users, to make sure the design will meet the needs of at least a large proportion of the users. Since I have only seen the most recent post from Jan, I don't know the full context of the discussion, but I would definitely encourage further research on the use of smoothly scrolling tactile display for both Braille and graphics. Best regards, John Roberts Project leader, NIST Braille / tactile graphic project john.roberts@nist.gov http://www.nist.gov/braille At 07:12 AM 3/11/02 -0500, you wrote: >This is quite interesting. I've ccd John Roberts who is working on the >nist wheel so that he can enter this duscussion since they've put a >great deal of research into this. I would also like to add that the >idea of horrizontal scrolling in a smooth way is not unappealing and I >understood it to be this instead of jumping from letter to letter. What >you want to do is to simulate the finger as much as possible and for >that, you need to do a bit of research. For instance, at some points >along a continuum, more examination is needed than at others which is >why it is hard to produce a scrolling effect that works reliably. The >other part of this is that with the horrizontal scrolling, it is only >necessary to show the left, middle and right side in rapid succession or >at whatever rate the user needs. I would recommend actually that the >character be devided up but that there be a deffinite break between >characters in the scroll so that this could help in character >deffinition. On the vertical, I have had no experience but would find >it difficult to orient myself to such a scroll. > >----- Original Message ----- >From: "zzzzzzzzzt" <jf.blinux@dynolink.org> >To: <blinux-list@redhat.com> >Sent: Monday, March 11, 2002 1:57 AM >Subject: smooth scrolling... braille output hardware (jf) (LONG) > > >Thanks everyone for the input provided so far. >Even the 'flames' provided data! (feel used?:) >This post is long,(you have been WARNED) > >Based on the responses to the issue of scrolling, I think there may be a > >partial mis-communication of my (sighted) scrolling concepts. >Please forgive me, but I have never before tried to explain anything >omitting any visual references. Its a new paradigm shift. > >Things I totally understand so far. >Any reader requires minimum of 2 fingers worth of data always available. > >The 2 finger positions if width adjustable(rigid/slideable) >should be incorporated if possible to improve ease of use. >Requires ability to easily jump around on the line for review etc. >Without those minimums usage would be miserable at best. > >The thing I think is mis-understood is the scrolling part. >I am talking about SMOOTH scrolling, not jumping a whole character or >even necessarily a whole column at a time. Many smaller in between jumps > >are displayed. The brain fills in the blanks reliably at some point. >Thats how movies,TV, and computer audio do it too. > >With a standard text LCD display driver chip... >Worst case with 1 braille character mapped to 1 ascii character would be > >only 5 steps of smoothness per full character position. This = 1 LCD >dot/1 braille dot ratio. >If each COLUMN of braille were mapped to 1 ascii character it would be >10 steps per character position. The ratios are flexible depending on >required space between dots. With a 1/2 ratio only 16 seperate column >patterns are needed(only 4 bit/dot positions). >5 steps may be too small for everyone but 10 might suffice. >If not, so much for using text LCD drivers for braille! >With a text LCD driver chip only 1 height is supported. >They are already capable of smooth scrolling a 5or10 column chr. > >With a 'graphics' LCD driver chip... >Imagine each DOT made up of 64 smaller dots in a 8x8 grid. >To move a dot just one whole braille column position would require 8+x >seperate sub-dot column shifts. The first 8 just to shift the current >dot away, and bring the possible next dot into center position >(the space between columns and chrs is flexible). >To move a set of dots a whole character position would produce a minimum > >of 16+x seperate interim dot feelings (x=additional space between >columns and characters). Put a caterpiller on your finger and feel the >sensation of ??? moving with their many legs as they walk. (yuk) >Circular dots are simulated if required by not raising the corner >sub-dots. The corner sub-dots are still needed because they will be >middle sub-dots as the dot is scrolled by sub-column. >With multi height per sub-dot round dot tops can be made, else flat top. > >Concave and other non round shapes are doable with multi height. >The difference between economy and elite now becomes the number of >sub-dots/dot over some acceptable/useable minimum. >Graphic LCD driver chips are available in B&W(single) and >greyscale(multiheight). > >The sighted analogy of a 1 character display provided was EXCELLENT. >If text moved smoothly into and out of view I could read accurately >slowly. >If each character was instantly replaced by the next I would have >trouble reading accurately even slowly. >My sighted and braille analogy is to cut two 1-2 character holes or in a > >piece of paper and read a line of text/braille with it by moving it >smoothly aover a line of text. If you had to close your eyes or lift >your fingers between the time each character lined up exactly, it would >be quite difficult to read I agree. (even if someone else was doing the >moving automatically) If you can sample even 1 or 2 times more between >characters centers it gets easier quickly. >A sound analogy is listening to someone speak normally, listening to >them speak from behind a slow and fast spinning window fan. Worst case >here >would be a fan blade made out of a solid disk with just 1 slot cut out >also spinning slowly. you would only hear the speech when the slot was >lined up with the speakers mouth. (jerk scroll) >Think window fan, and you cant hear around the fan, because of the house > >wall, only between the fan blades. >Different fan speeds represent different resolutions of sub-dots. >Each sounds different and provide different amounts of data and the same > >words. At some point the words are lost representing minimum useable >resolution. >The PC speaker/piezo can make OK understandable words with only about >1/8 >the samples considered minimum for audio! > >A simple thin soft flexible covering on the whole thing would further >'average' the heights during and between steps, and enhance overall >response. Consider this the only possible consumable necessary besides >power. Similar averaging is responsable for the success of audio >digitization and all lossy compression. (results vary) > >I have a few solutions to the jumping around the line. >Put wheels on the bottom of the 2 finger reader and run it back and >forth in a simple hot wheel track. A pot attached to a wheel does the >mech/electro part and the SW the rest. (many variations possible) >Put a thumb operated slider pot to indicate where in the line the >fingers are desired and the same SW as above. (most portable idea) >The slider can be placed upside down under the finger positions for >better ergonomics. It would be held between the thumb and fingers IN the > >hand. > >To convert one of theese to 'graphic' start mid-res and scroll whatever >in x and y instead of just x. The multi sub-dots already puts it in the >graphic league! Add more fingers too! >The software needed is not exotic at all. > >Again a portable model would use thumb positioner within a smaller frame > >and just have small thumb movement equal large 'screen' movement. >This also easily ergos into an IN hand model. >This is how a mouse user moves a pointer everywhere on a big screen on a > >smaller mouse pad in one movement. The acceleration becomes unnoticeable > >when adjusted properly. The touchpad on my laptop is way tiny but it >works! > >To read text on such a 'screen' with linux it is just a matter of >selecting braille as the display font in the size desired. > >Questions follow: >Is my description of smooth scrolling comprehensable to the non-sighted? > >Is my concept of SMOOTH scrolling different that your concept of text >scrolling? >Is SMOOTH scrolling what was described that works only so/so now??? >If one existed does it sound useable? > >That it cant be done I do not accept, I am incapable. >That it wont be done can be arranged. >I dont claim to be the best man for the job either just willing. >SOMEONE ought to define whats needed in minute detail and isolate the >engineering obstacles remaining. >Every USER will need to understand same anyways at first for sure!!! > >Jan P. Finegan >jf.blinux@dynolink.org > > > >_______________________________________________ > >Blinux-list@redhat.com >https://listman.redhat.com/mailman/listinfo/blinux-list