#5202 NORM Future : Acoustic Tape Measure thinks I'm short!

[Zarro Boogs per Child]

#5202: Acoustic Tape Measure thinks I'm short!
----------------------------------------+-----------------------------------
  Reporter:  cscott                     |       Owner:  bemasc        
      Type:  defect                     |      Status:  new           
  Priority:  normal                     |   Milestone:  Future Release
 Component:  acoustic-measure-activity  |     Version:                
Resolution:                             |    Keywords:                
  Verified:  0                          |  
----------------------------------------+-----------------------------------

Comment(by ohshima):

 Replying to [comment:6 bemasc]:
 > Replying to [comment:5 ohshima]:
 > > Showing all measuremant is just an attempt to obscure the fact that
 this is not the best tool to measure the 1-2 meters distance.
 >
 > A meter stick would definitely be better, but I don't know if they will
 have meter sticks.  Maybe they can make some.

 Maybe?  Again, if they have an XO, its dimension is known so they can for
 sure be able to measure stuff with it and make a meter stick, and a thread
 and folding it will let them put the fractions on it.

 > I agree.  I am open to suggestions.  However, it is very difficult to do
 better than "look at the code".
 >
 > The activity relies on special autocorrelation properties of maximum
 length sequences produced by linear feedback shift registers.  It uses
 FFT-accelerated cross-correlations of these sequences to determine the
 time-delay between the two laptops.  There's also some very tricky timing
 logic required, because it is impossible to synchronize the clocks to
 sufficient accuracy over the network.  What I'm trying to say is, I have a
 degree in mathematical computer science, I wrote the program, and I still
 don't really understand why it works.  I have no idea how to explain it to
 a student, even an exceptionally bright 16-year-old.

 No.  You have to give them a right abstraction.  The explanation on the
 wiki.laptop.org doesn't mention the detail but almost understandable for
 kids.

 For example, random white noise gives you a signal whose energy is spread
 over wide frequency range.  If you look at it in a sonogram (like the one
 in Etoys), you get a black line so it is easy to detect it.  If the frame
 size (# of samples) is 32, you would still get 30cm accuracy with 44.1kHz
 sound.  That is ok for a distance in 10-20m range.  That version is easier
 to explain in a honest way for children.  We could create an Etoy (for
 example) and they could really see the sonogram visualy and mess up with
 code on XO while it is running.  (Measure activity and Acoustic Tape
 Measure should be able to *use* other.)

 I and my colleagues talked about it and it would be even better to be able
 to specify different sound.  That way, kids will try to intervene the
 friends measurement, etc. and they will think different ways to make it
 robust and some will "discover" that white noise is pretty ok.  (No kids
 will "discover" the maximum sequence length thingy, but that is ok.)

 Also, making a simulation is another idea.  After doing the real
 experiment, kids can locate their simulated laptops on screen, and write
 the essence of simulation ((t1-t2)/2/s).  They can change parameters, etc.
 This would be constructive.  (The "active essay" idea).

 We definitely want to have a way to use the distance readout in some other
 activities that can do graph, etc.

 > > and the UI should be simpler.
 >
 > How could the UI be any simpler?  It's got one button and one display
 field.

 I was more comparing it with the "four readout" idea.  But the parameter
 setting can be always visible, for example.  So kids don't have to switch
 tabs.

 > > (Probably response time should be a slider to change the interval,
 though.)
 >
 > This might be a very good idea.  It would not be very difficult to add a
 slider to change the speed.  However, it would add complexity to the
 interface.  Alternatively, I have considered a system that automatically
 adjusts the speed depending on distance and background noise (the lower
 the SNR, and the longer the sequences must be to make correct
 measurements).

 It can be semi-automatic, yes.

 I can't help wondering if you have read my email to devel mailing list:
 http://lists.laptop.org/pipermail/devel/2007-November/007805.html

-- 
Ticket URL: <http://dev.laptop.org/ticket/5202#comment:7>
One Laptop Per Child <http://dev.laptop.org>
OLPC bug tracking system