Subject:
|
Re: The least you could do!
|
Newsgroups:
|
lugnet.org.ca.rtltoronto
|
Date:
|
Mon, 25 Feb 2002 03:57:56 GMT
|
Viewed:
|
535 times
|
| |
|
|
In lugnet.org.ca.rtltoronto, Ivan Louch writes:
> I'm really surprised that no one else tried to implement something similar.
> Except maybe for Bruce - appearently his bot could lift up two blocks.
We thought about this for a while, but decided against it, for this reason:
If you run a couple of the possibilities through, the magazine is more
useful with a large board where there's a big distance between where blocks
are and where they're needed. Meaning, assume the board was 10x10 instead
of 4x4. Say you need to place a block at position (2,10) and the next three
blocks adjacent and there's a string of blocks at (0,{1,2,3,4}) location at
the start. The magazine then saves you four trips back and fourth a
relatively long distance, which also should reduce your accuracy problems
too, because you're not going back and forth the entire distance of the board.
Pre scanning the board like Chris is a similar strategy that works well for
a board that is much larger but doesn't buy a particularly strong advantage.
If we had a 10x10 board and only one block was different than the target,
Chris could scan the entire board and change the one block versus our code,
which goes sequentially and fixes on the fly.
The problem is that there aren't enough spaces/blocks/moves in the 4x4. The
worst case scenario is the AntiX pattern, and that's eight full moves. But
the majority of the time in the game is gripping and dropping, versus
movement, which you've got to do with the magazine anyways. The AntiX
doesn't require the magazine to fill. Even if all the blocks are at the
close end, you only need to move four of them, and the four which need to
move aren't moving so far to give a big speed saving.
> Even though I expected a bit higher scoring I'm still very happy that my bot
> worked at all, since it is my most difficult project yet. I just wish I had
> more time for tweaking and debugging, as my biggest problem certainly was
> unreliability due to lack of play-testing (mainly the software that I wrote
> some time between Friday evening and Saturday morning) . Must say I greatly
> underestimated the complexity of the task, so let us learn on out mistakes
> :))
I'm super impressed of folks like you, John, Matthias and Tony (who actually
has been to several rtlToronto events, often building for them, quietly) to
take up a challenge like this. Agreed, good show!
Calum
|
|
Message is in Reply To:
 | | Re: The least you could do!
|
| > Come on guys! Those of us who lurk at your news group are dying to know who > won this battle of the bots. Was it Rob again? Another battery box wonder? Well, it certainly had a number of battery boxes, as a matter of fact Rob had as many battery (...) (23 years ago, 25-Feb-02, to lugnet.org.ca.rtltoronto)
|
11 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
