Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!fas-news.harvard.edu!newspump.wustl.edu!news.starnet.net!news.starnet.net!news.dra.com!hunter.premier.net!news.maxwell.syr.edu!cam-news-hub1.bbnplanet.com!news.bbnplanet.com!howland.erols.net!newspump.sol.net!newsfeeds.sol.net!news.webspan.net!news.tacom.army.mil!ulowell.uml.edu!news.wizvax.net!mv!usenet
From: hodsdon@scoot.netis.com (Steve Hodsdon)
Subject: Re: Multiple-Choice best-path algorithm?
X-Newsreader: Anawave Gravity v1.10.556
Organization: Pragmatic Designs
Message-ID: <MPG.d6eb3d48ba1f2cb989684@news.mv.com>
References: <5d6fec$guh$1@missing.link.ca> <MPG.d6d13ae2243b396989685@timrgate> <5e03jh$beg@chronicle.concentric.net> <MPG.d6e4f8d85d7e904989688@timrgate> <5e25t4$p7d@dfw-ixnews5.ix.netcom.com>
X-Nntp-Posting-Host: scoot.netis.com
Date: Sat, 15 Feb 1997 02:36:34 GMT
Lines: 75

In article <5e25t4$p7d@dfw-ixnews5.ix.netcom.com>, edybs@ix.netcom.com 
says...
> In article <MPG.d6e4f8d85d7e904989688@timrgate>,
> 	hodsdon@no-spam-please.scoot.netis.com (Steve Hodsdon) wrote:
> 
> >
> >This brings up a (perhaps) stupid question.  If my, (the human player), 
> >unit moves from a hidden location to a visible location then back to a 
> >hidden one, does the AIP "notice" this?
> 
> Good question, and I think your question further suggests a broader question:
> 
>   "Should the path finder just find the best path, or should it perform
>    threat assessment, goal assessement or other non-pathfinding-specific
>    AI functions, while it is in the process of finding the best path?"

I'm going to put myself on the spot here and say that the path finder 
should only use the terrain features when run.

[...]
> other stuff).  Maybe, this is not such a great idea?  Maybe these
> non-pathfinding-specific functions should be a part of pathfinding?
> 
> What do you, and others think?
> 

I think that the map should have the "threat" level encoded into it.  
(For an old discussion on this, look for the influence thread on Steven 
Woodcock's AI page.)

> 
> >
> >I'm beginning to wonder how much CPU time it's going to take to calculate 
> >paths for each unit...  Something I don't worry about with a test program.  :-\
> >
> 
> If I may suggest, that in a real-time game like C&C/RA/WC/EN that the speed
> of pathfinding is critical.  Consider that with 2 players with 40+ units

I have to agree here.  

All my energy has been directed at turn based games.  The AIP can use as 
much time, within reason, to plan its next move.

Now, correct me if I'm wrong.  I can see two basic methods for path 
finding.  One, which should be the fastest, is to find the straight line 
distance from point A to point B, using Bresenham's line algorithm.  The 
other would be to use A* or Dijkstra's algorithm, (or some other one).  
We can complicate (read slow down) the algo by considering other details 
such as the cost of the terrain or "threat level."

As an aside, I can't see re-calculating the path after every move.  (I've 
seen games where a unit will bounce back and forth between two hexes.) 
I'd druther just re-calc if I get stuck.

Back in the influence mapping days I thought I would look at the six 
surrounding hexes and move towards the strongest value.  This way, my 
unit would move to intercept the enemy.  Now I realize that there could 
be no "combined action" options available.  Units would not work 
together.

Now I'm looking at a leader requesting a path to a target and then making 
the resulting data available to it's subordinate units.  (Let's say that 
there is four units in a platoon, the CPU time for path finding is 1/4 
what it would have been.

[...]

Steve
-- 

I eat my peas with honey.  I've done it all my life.
It makes my peas taste funny.  But it keeps them on my knife.

mailto:hodsdon@scoot.netis.com
