|
[DPRG] Sonar 1/r (not 1/r squared)
Subject: [DPRG] Sonar 1/r (not 1/r squared)
From: Chris Jang
cjang at ix.netcom.com
Date: Thu Jun 14 16:04:28 CDT 2007
>And on to other things: The next thing I want to do with sound is
>to make a sound-driven weather station, to give temperature,
>humidity, pressure, and velocity using nothing but three
>transducers (no servos). Temperature and humidity affect the
>speed of sound. Pressure affects the intensity. Velocity affects
>the arrival time. What I haven't figured out is how to separate
>the humidity measurement from the temperature measurement. And
>what about measuring rainfall?
I'm ignorant of the physics involved in this problem so am only
speculating in the abstract.
Perhaps the relationship between these hidden state variables is
multimodal? Then a statistical approach is motivated, probably
with something like a particle filter. The combination of
temperature, humidity, pressure and speed might be represented as
a probability density in several dimensions (four or less) over
time.
And if we assume that this is not the case, then I offer that there
is still uncertainty as we are sampling and inferring the hidden
state. So that leads to something like a Kalman filter.
More information about the DPRG mailing list
|
