BurntFingers
Posts: 155
Joined: 6/28/2004
Status: offline
|
quote:
ORIGINAL: VCDH
First, I'd like to say that these ideas are really good. Please keep them coming.
Second, a rough statment regarding the ultimate incarnation of what we plan is an AI that can dynamically respond to a tactical situtation that it can analyze on it's own. So while the player would be tasked with *do this* the AI would recieve a rough idea of it's own tactical situtation, i.e. *protect this*.
It sounds to me very much like the AI managing to interpret the Victory Conditions and use Mission Planning/Formation Editing/manipulation of naval units to achieve Victory Conditions.
This isn't quite what the original post was about - that was about using a Scripting Language to modify behaviour.
Now, my 2 cents worth;-
1) The main sticking point is an AI that can intelligently allocate assets to missions. That's the real sticking point, and Python just won't cut it for that (too slow, too many variables). But what I reckon you could do is have an AI that can allocate assets to a SINGLE set of Victory Conditions. It's not an easy job but it is just about possible within current known game theory/PC processing power.
2) What the Scenario Editor could have is choose what sort of "bent" or inclination the AI uses in acheiving Victory Conditions. For instance, there could be a set of switches ranging from Fastest Solution (Lots of Air Power, lots of PGM), Sneakiest (minimal use of active sensors), and settings for agression/defence aspects.
That way a Scenario Editor would have the choice of setting up an AI opponent.
Alternatively, have them done as a set of ingame choice - additional screen to selecting difficulty level (it would actually REPLACE current difficulty level).
One thing the miniatures game has that computer Harpoon doesn't is pilot ability - novice, regular, veteran, ace etc. That's the sort of thing AGSI should be thinking about implementing.
I can waffle on about this for days. Got to go now, I'll think about it some more.
On a lighter note; boyz with batz with nailz through da ends. ;)
This will ultimately lead to the AI targeting the amphibious force as opposed to the carrier because the amphibious force is what can put men on the ground in the area the AI is protecting.
The bad news is that before we can do that, we need to decouple other components from the game engine to make them more modular. We'd also like to make the AI programmable to a certain extent, not using Python or anything like that, but more of what you see in a Steel Panthers kind of Preferences Menu, just tailored for naval warfare.
Later
D
|
Printable Version
Future of Harpoon - 
) would perhaps 'cheaper' to build as a sophisticated scenario/mission editor. 
. The problem with the approach which is currently available in Dangerous Waters/Steel Beasts Pro is that you can somehow modify the behavior of the AI (in DW you have doctrine files in addition) but it's very painful to do so. Actually, you define rules/constraints and since they have dependencies to each other, it's very difficult to keep the overview when you've a lot of rules. 
New Messages
No New Messages
Hot Topic w/ New Messages
Hot Topic w/o New Messages
Locked w/ New Messages
Locked w/o New Messages
Post New Thread