Subject:
|
Re: Code Repository
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Thu, 16 Sep 1999 17:29:53 GMT
|
Original-From:
|
Jonathan Knudsen <jonathan@oreilly#NoSpam#.com>
|
Viewed:
|
502 times
|
| |
|
|
At 09:35 AM 9/16/99 -0700, alex wetmore wrote:
> [...]
> I think it is even interesting to see how different people attack a basic
> problem, such as the line following algorithm. While reading the chapter in
> the O'Reilly book on NQC I noticed that their line follower has a much
> different algorithm then my own. They would go back and forth a large
> amount, but always go the same amount. Additionally, they always started by
> going left. My script would sweep left and right a small amount, and then
> grow that amount until it found the line. It also remembers which direction
> it found the line in last time, and searchs in that same direction again.
Actually, it sounds like our algorithms are more similar than
you think. My assumption was that the robot would cross the line in
a zig-zag motion, driving off the line on alternate sides. Based
on this assumption, I remember the last direction turned and
always try to turn back to the line. If it times out without
reaching the line, I turn back the other way. As you've observed,
the track shape determines the success of the algorithm. On
an oval track, for example, the robot tends to drive off the
line on the same side each time, and my algorithm is not
particularly efficient in this case. (I described the algorithm
better in Chapter 3; in Chapter 4, I just re-implement it in NQC.)
Jonathan
|
|
Message is in Reply To:
 | | Re: Code Repository
|
| (...) I'm completely in favour of something like this. Now I wonder what's the best way of implementing it, because HTML doesn't seem to be the best for copy-paste... Laurentino Martins [ mailto:lau@mail.telepac.pt ] [ (URL) ] ---...--- “For once, (...) (25 years ago, 16-Sep-99, to lugnet.robotics)
|
17 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
