Subject:
|
Re: THOUGHTS on mapping with Lego Robot ( added to searching walls)
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Tue, 27 May 2003 23:18:56 GMT
|
Original-From:
|
Steve Baker <SJBAKER1@nospamAIRMAIL.NET>
|
Viewed:
|
931 times
|
| |
|
|
Bill Blackmer wrote:
> What I want to do is take a five foot by five foot square, put random items
> in the square, MAP THE SQUARE AND OBSTACLES and run our dog robot around the
> course and never have it stop. I do not want to do obstacle avoidance (that
> is much easier) I want to build a map and work from it.
It's very difficult because all of the 'dead reconing' methods out there are
approximate and accumulate errors.
If you drive your robot 10 feet in a straight line, your software will probably
by 'off' by maybe a half inch, after 20 feet, you'll be off by an inch...after
an hour of driving around, you could be just about anywhere! Turning is
worse still - if you drive around a zig-zag course for 10 feet, you'll probably
be off by many inches in position and tens of degrees in heading.
You need some way for the robot to (at least periodically) do a check on
some absolute position to reset it's internal position indicator.
> To do this, we have acquired several HiTechnic Sensors (these sensors are
> much more accurate then the Lego sensors - Compass, Ultrasonic and Infrared
> Range Sensors); and we have created several statistical averaging and
> correction algorithms, etc.
But things like wheel slip, slop in the gear train, small errors in the
diameters of wheels, the fact that the Lego rotation sensors sometimes
gain or lose a 'count' (especially when moving slowly)...all of these
things conspire to make this a completely intractable problem.
There are things you can do to lessen the problems - but *some* inaccuracy
is certain to occur and if the "dog" robot doesn't periodically know with
some precision exactly where it is, the errors will build up with alarming
speed (ESPECIALLY when the robot turns).
One idea that might work would be to have a 'kennel' for the dog - with
a funnel-like entrance and an interior that *exactly* fits the shape of
the robot. If the dog returned to the kennel every so often, the funnel
would point it into the kennel even if it's navigation was off by a bit
and the tight fit inside the kennel would allow the robot to reposition
it's internal navigation coordinates to be EXACTLY where the 'kennel'
is known to be. There are other ways to achieve the same effect - a
barcode placed on the floor, navigation lights placed around the arena,
etc.
---------------------------- Steve Baker -------------------------
HomeEmail: <sjbaker1@airmail.net> WorkEmail: <sjbaker@link.com>
HomePage : http://www.sjbaker.org
Projects : http://plib.sf.net http://tuxaqfh.sf.net
http://tuxkart.sf.net http://prettypoly.sf.net
-----BEGIN GEEK CODE BLOCK-----
GCS d-- s:+ a+ C++++$ UL+++$ P--- L++++$ E--- W+++ N o+ K? w--- !O M- V-- PS++ PE- Y-- PGP-- t+ 5 X R+++ tv b++ DI++ D G+ e++ h--(-) r+++ y++++
-----END GEEK CODE BLOCK-----
|
|
Message has 5 Replies:
Message is in Reply To:
12 Messages in This Thread:
   
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
