DPRG
DPRG List  



DPRG: Neural Network(Yes!) robot revised

Subject: DPRG: Neural Network(Yes!) robot revised
From: Corey Hansen nzhansen at ihug.co.nz
Date: Mon Jan 12 20:45:07 CST 1998

Hi,

Alright , here's what I have on the issue of my robot. 8 inputs; 3 IR , 4
bumper switches , and a Cds cell. The robot compiles all the inputs so they
are represented by an 8 bit variable. The robot then goes to the address in
the code which corresponds to the sensor inputs. Now the robot needs to
decide on the motor output remembering the values of the 7 variables
representing the outputs. Now that the robot has the output figured out it
starts it. Now it waits until a sensor input(s) change(s). Now it compares
the older input with the new. If the input is lower then before then the
robot upgrades the  previously used output degrades the others  , or vice
versa if the input is higher. Then the robot repeats this , learning the
best use of outputs of the most common inputs.

Another thing is the need for long term learning. For this the robot needs
power for a long period. I could tether it but , not as neat. So I'm going
to try to make a charging station for it. I'm going to have a low voltage
detector connected before the regulator to detect when the batteries get
low.
The output of this is connected to the XIRQ pin to alert the robot. Then the
robot needs to search for the station. This makes use of the front facing IR
sensor and the Cds cell. The robot , when alerted , will turn in place until
it sees an IR signal with it's sensor. This front facing sensor is set to
receive 38khz modulated IR so its range is about 20 feet(According to my
testing) If the robot doesn't see the station when it has turned a few times
then it will go back into normal operation for a minute or so to get into a
new position and repeat the looking process.

When(If) it does see the station it will head in its direction. The only way
to know when it is near is the use of the Cds cell which will be facing
down. During normal operation it will be treated as a bumper switch. But ,
during low battery operation it'll detect white paper that will surround the
charging station. Now that it knows it's in the charging area it'll just
head forward until it hits something. When it hits something it will now
know it's ready to charge because on the end of the bumper switches will be
the charging probes. The charging station will be circular so that it
doesn't matter what angle of impact the robot came at , and there will be
tin foil pads properly spaced for the probes. I'll include a rectifier
bridge to protect against reverse polarity.


What's coming through the probes will be 7 or so volts , and the bridge will
drop that a bit. What's connected here will be connections to the motor
battery , processor power pins , and the input to the voltage regulator. In
between the voltage regulator and the power pins will be 1 or 2 darlington
transistors. These darlington's will be used to isolate the  battery power
so it can charge. These transistor will be controlled by the micro. Oh , and
the motors need to be stopped so that their batteries will charge. Now I'll
have the robot sit there for a certain amount of time , and then it will
turn on the darlington's  , and motor backwards UNTIL it sees , with the Cds
cell , that it's cleared the white paper. Now it goes back into normal
operation.

Notice the large use of the future tense!

Any suggestions would help , a lot.

:David
davehansen at geocities.com

------------------------------

More information about the DPRG mailing list