Get Rid Of Ant Petition

[Learner]

For players it is at the individual level.

 

Currently, the decision is made by each Ant, but they do have a swarming characteristic already when an Ant gets hurt and others are nearby. But with their levels that currently means the player will just kill the other Ants. They just don't initiate a swarm, yet

[Learner]

 

How long have the ant's been back in action.....a week or so?

161281[/snapback]

They still have their old memories. So people that attacted them before, the Ants still remember. But, they haven't learned you've leveled.

[Earl]

Is it possible to eventually give monsters this swarming property? If you have 21 defence a goblin would never attack you, however if a goblin sees you already engaged in battle it might want to join in. Might make things more interesting.

[Deson]

For players it is at the individual level.

 

Currently, the decision is made by each Ant, but they do have a swarming characteristic already when an Ant gets hurt and others are nearby. But with their levels that currently means the player will just kill the other Ants. They just don't initiate a swarm, yet

161287[/snapback]

 

Ok, so If I understand it correctly, they are already setup to swarm, but don't becasue of their levels at this point. But once their levels go up then that maybe another matter then.

 

Should prove a interesting time when that happens, if it does.

[ttlanhil]

I think someone should put a notice up...

 

Seeing an ant is not reason to file a 'bug' report

 

because I reckon someone will do that eventually

[xeLJoYo]

This seems to be a truly amazing AI...

Totally new indeed. U could make money out of this

[duran]

while it have a lot of data to crunch it looks no more advanced then a chess computer with some dynamic bits added in.

 

i recall running into something similar in a game i have, after every round of game a message would pop up saying that the overlord was learning. basicly i think it was crunching what i was doing the whole game and trying to look for patterns and how to mess with said patterns.

 

ie, if i allways goes for nearby resources and create few but powerfull monsters then it may counter by denying me said resources and going for some way to take down my monsters the best way.

 

still, i would love to see the say a ant would run from a bear as it knows it cant win against it

[Learner]

while it have a lot of data to crunch it looks no more advanced then a chess computer with some dynamic bits added in.

 

i recall running into something similar in a game i have, after every round of game a message would pop up saying that the overlord was learning. basicly i think it was crunching what i was doing the whole game and trying to look for patterns and how to mess with said patterns.

 

ie, if i allways goes for nearby resources and create few but powerfull monsters then it may counter by denying me said resources and going for some way to take down my monsters the best way.

 

still, i would love to see the say a ant would run from a bear as it knows it cant win against it

162940[/snapback]

They don't run away currently, but will tend to ignore the bear for a long time, but they currently are checking too often, so even having only a 5% to win, if the bear sticks around very long, eventually the Ant_Queen does attack

 

Yes, they need the concept of fear. First, I need to set them up so they can learn what monsters are willing to attack them at what 'level' they are. Then that combined with the chance to win can help with the fear and running decision.

[Tanyia]

Swarms of ants? Scary...anyone read "Prey" by Michael Crichton?

[Guest Bombinette]

FYI, they know about 16 maps currently, and WS alone requires 28M. I'll have to make the map loading more dynamic before the second continent is released or my memory usage will double.

161256[/snapback]

 

Just an idea : could you save memory space by using a kind of data compression ? Maps have most of the time large areas with identical values, so using compression may be considered. Or you could store only the coordinates of the features from a map, all other coordinates considered as walkable and ant paths being new features.

 

 

PS : I love these ants !

Edited May 18, 2005 by Bombinette

[Learner]

Just an idea : could you save memory space by using a kind of data compression ? Maps have most of the time large areas with identical values, so using compression may be considered. Or you could store only the coordinates of the features from a map, all other coordinates considered as walkable and ant paths being new features.

PS : I love these ants !

163073[/snapback]

The data I'm tracking as part of this memory usage is Ant specific data only. What increases memory usage currently is the fact that there are 16 different types of trails that a single spot can have and 4 of those can have additional information (info about specific harvested items). Walkable takes 1 bit, an item being harvestable at that location is another bit, same with other data. Currently I'm only using 10 of the 32 bits in the flags section though.

 

I do keep a portion of the real map data in memory, and that is stored in a fairly compact format.

 

But, work on the LRU system with map data will help by dropping the detailed map data for maps an Ant isn't on or hasn't been near currently. The been near is needed since when they are evaluating a Tport/map change they need to get some data from the destination map.