anelok user interface
sadov at linux-ink.ru
Wed Sep 18 18:19:35 EDT 2013
Example of realization of keyboard for small touch screens -- Carnegie
On Thu, Sep 19, 2013 at 1:06 AM, Werner Almesberger
<werner at almesberger.net> wrote:
> Felix wrote:
>> I don't know if Werner have something in mind,
> Of course :-)
>> 1. Insert pin screen
>> You have to insert a numeric pin of some fixed length.
> Not sure if a number is ideal. Letters may work better because
> they're easier to remember. (More below.)
> I was thinking of the following input scheme:
> - rotate in any direction to get the first letter/digit,
> - rotate in the opposite direction for the second,
> - change direction again for the third, etc.
> - to delete one letter, press the center button, after which
> you turn in the direction you want.
> - waiting 0.5 s (or similar) ends input and starts validation.
> The display would show the current letter/digit plus briefly
> the last one, so one could be sure that the button press didn't
> alter it.
> The display has room for about three lines of reasonably large
> text (~12 pt, bold and illuminated). During code entry, one line
> could show the alphabet in a tiny (4 pixels wide) font. That
> would be hard to read but all you really need to see is the
> rough shape of letter. There would also be room for making some
> of the letters a little larger.
> Have a little bar move below the letter to indicate the wheel's
> "position" in the alphabet. That would allow to estimate the
> distance to the next letter and "aim" better.
> I came up with the following equations:
> a = alphabet size, 10 for digits, 26 for letters, etc.
> e = maximum entropy per letter, in bits
> p = average number of positions from one to the next letter
> u = difficulty of rotating the wheel by one position
> r = difficulty of reversing the direction, in units of "u"
> m = difficulty of remembering one letter, in units of "u"
> t = total difficulty of entering one letter
> E(a) = efficiency of an encoding, in bits per difficulty
> e = log2(a)
> p = a/2
> t = p*u + r + m
> E(a) = e / t
> We can rewrite this to N(a) = log2(a) / (a + C) where N(a) is a
> "normalized" efficiency (we don't care about the precise value)
> and C is the per-letter difficulty in relation to moving the
> wheel by a certain angle.
> Now, to find the maximum, we can either get out pencil, paper, and
> our old calculus textbooks, play with www.wolframalpha.com (*), or
> simply plot N(a) for various values of C (**) and make an eyeball
> (*) E.g., for C = 20, a possible query would be
> maximize log2(a)/(a+20)
> (**) E.g., for C = 20, and dropping the constant factor of ln(2),
> plot [a=1:30] log(a)/(a+20)
> Now assign difficulty values, substitute "maximum" entropy with
> real-life entropy, and run the math :-)
>> 2. Main screen
>> You can scroll up or down among the names of stored data
>> 3. You choose an entry and see the data, scrolling through info. Long
>> press takes you to main list:
> That, or simply have a "Back" item in the list or make the title
> act like "Back". I like the general idea of using the length of a
> button press, though. That's something that's often overlooked.
> We may reserve it for administrative functions, though.
>> 4. If you leave the device on without interaction for more than 2 mins
>> (e.g), the screen will be powered off and the cpu will go to deep sleep.
> I'm thinking of the following states (among other, more specialized
> - off: minimum power consumption, password safe is locked,
> pressing the button a brief moment (longer than just hitting it,
> to reduce accidental activation), turn on and show the PIN
> - standby: display is off, LED is lit (possibly via PWM, to save
> power), password safe is open. Switches to "off" after a while
> or when turning the wheel. Goes to "open" when pressing the button.
> - PIN dialog: as described above. With automatic return to "off" if
> left idle for more than a couple of seconds.
> - open: display is on, password safe is open. Goes to "standby" if
> idle for a while.
>> Well, very basic, but is a idea to start, no?
> Yup, we're thinking in the same direction :)
> - Werner
> Qi Hardware Discussion List
> Mail to list (members only): discussion at lists.en.qi-hardware.com
> Subscribe or Unsubscribe: http://lists.en.qi-hardware.com/mailman/listinfo/discussion
More information about the discussion