Subject:
|
Re: Brute Force Brick
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Tue, 3 Feb 2004 23:01:59 GMT
|
Viewed:
|
1007 times
|
| |
|
|
In lugnet.robotics, John Barnes wrote:
> I have always listened, with great interest, to all the various threads that
> pop up on this subject. Its one of those subjects which compells many people
> to express their opinions and also seems to attract almost as many different
> opinions.
Let us enjoy some philosophics:
Why use a numbercrunching brute force brick on a simple bot control application?
As this question implies, this makes no sense. Sheer computing power is
required, if you use intelligent software, than can determine the position of
the robot in a given environment simply from its, for example, distance
measuring sensors. Such that slipping of the robot wheels on the ground does not
matter.
Sensor Fusion in Certainty Grids for Mobile Robots
Hans P. Moravec / AI Magazine / Summer 1988
Hans P. Moravec from Carnegie-Mellon University once wrote an article that
solved this problem basically (You get it in any good computer science library).
As i think, in this sense, he is the father of the spirit and opportunity rovers
as well, as i think, his technique is used as a base technology in all those
rovers.
This is, as i think (but i dont know, i only suggest) because of the mathematics
he uses to solve the problem. His Bayesian certainty calculation resulting in a
certianty grid, in which, to speack in practical terms, the robot can move,
allows to relinquish from a potential wheel distance measurement information.
The Bayesian certainty calculation requires brute force, indeed!
I'm thinking about the bots that we could build, if we had a software
system - however it might look - to write contol programs basing on certainty
grids.
Imagine a robot vaccuum cleaner.
First he has to examine, learn the room, then he has to clean it.
I am thinking of the Bayesian certainty calculation as a software subsystem
used by a User Written Application System. First, we must create an interface.
The Bayesian certainty calculation could be a part of the brute force machine,
it should be shipped with it.
Probably, it would be a good idea, to make it a GNU open source project, such,
that the software will be free and available without costs to all people.
After we had solved that navigation problem (its a control problem, basically,
and a software structure problem too), the next step, technology might take is
to navigate by image data, as the mars rovers do.
So now you got my opinion clearly: a brute force machine only makes sense, if it
has got some very intelligent software on board, using that brute force, for
intelligent purpose. Leaving the rest of its power for control application.
Greetings
Ralph
|
|
1 Message in This Thread:
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
