|
[DPRG] Re: Allen Robots (Will's emotions)
Subject: [DPRG] Re: Allen Robots (Will's emotions)
From: Randy M. Dumse
rmd at newmicros.com
Date: Fri Mar 9 13:37:49 CST 2007
John Swindle said:
> What difference does it make how the robot is programmed? It
> just does exactly what you told it to do.
But that's what is so surprising about emergence.
If I program the robot to do exactly this. Then I program the
robot to do exactly that. And when I test it, it does, this,
that, and something else. The "emergence" of the unexpected
something else is what emergence is about. W. Grey Walter noted
in the '50's his turtles were delightfully unpredictable. There
were couplings he didn't consider that gave unexpected
behaviors. The one that comes to mind had to do with his photo
sensor, which when the motors were running got less power, and
changed sensitivity. Which way I don't now recall. But none the
less, he would start the motors, and the light sensor would act
differently than they did with the motors off. Two significantly
different behaviors based on what one would have originally
thought were completely independent.
So while I agree perhaps a digital microcontroller does exactly
what you program it to do, or its broken, it is not necessarily
the case a larger system such as a robot will do exactly as it
is programmed. The program doesn't control everything in the
system. For example, the internal series resistance in the
batteries cannot be programmed. Just applying power to something
does not mean the power is applied with microprocessor
crispness.
For another example, you can tell a RC Servo to take a position
in ~1ms. That, however, doesn't mean it will be there in 1mS. Or
maybe even 100mS. Or ever. So the level of certainty we come to
expect inside our digital chips, has no real parallel in
physics-driven, inertia laden, macro analog, and micro quantum
"real world" out there.
The very point is: A robot does not do exactly what you program
it to do. It does exactly what physical laws allow it to do. Not
a spec more.
There is no firm causal link between code and action. There is
only a causal link between code and activation. At the level of
the robot, the code must respect reality, because reality has no
respect for the code.
Randy
www.newmicros.com
More information about the DPRG mailing list
|
