Currently Online

Latest Posts

Topic: News about AI

hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2017-11-18, 23:48

GunChleoc wrote:

I just had 4 AIs play on Crossing the Horizon. I watched one of them take take down a border forest in order to expand, well done face-smile.png

Only 1 of them built a port though and then didn't use its ship to explore. I guess we need more training on seafaring maps.

I just tried a couple of times to train Ai on the crossing the horizon map. Started with empire got promising results. nearly all AI did explore and colonize. After that tried it with babarians. realized that they couldn't build a port , cause they had no iron left. Added some iron to the map and eventually got the babarians to build ports and explore. but with these values empire did not explore anymore. will keep trying.
Another issue I think I saw was that the precious port spaces were sometimes just thrown away for other buildings. so I don't know whether the Ai is considering the port building lots as something special. so if tibor please could explain how the normally very rare port spaces are identified and treated by the AI. In my opinion AI need to know where to find them and if they are owned then they should not be wasted unless for a very good reason (e.g. basic economy not established, some ware absolutely necessary)
will keep you informed.


Top Quote
Tibor

Topic Opener
Joined: 2009-03-23, 23:24
Posts: 1377
Ranking
One Elder of Players
Location: Slovakia
Posted at: 2017-11-19, 10:44

hessenfarmer wrote:

Another issue I think I saw was that the precious port spaces were sometimes just thrown away for other buildings. so I don't know whether the Ai is considering the port building lots as something special. so if tibor please could explain how the normally very rare port spaces are identified and treated by the AI. In my opinion AI need to know where to find them and if they are owned then they should not be wasted unless for a very good reason (e.g. basic economy not established, some ware absolutely necessary)

One part of problem is that AI really does not consider port space until it owns it - similarly like human player until he sees the port building placeholder. So you probably mean that AI should cheat here a bit and just scan the map for all port spaces.....


Top Quote
GunChleoc
Avatar
Joined: 2013-10-07, 15:56
Posts: 3324
Ranking
One Elder of Players
Location: RenderedRect
Posted at: 2017-11-19, 16:42

I think cheating is OK here - just like with mountain resources.


Busy indexing nil values

Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2017-11-19, 21:55

In my opinion the Ai should treat the port spaces different from other fields which seems to be true if it is trained properly but in case there are only a small amount of them the Ai should know where to search for them. If you call it a cheat well the Ai deserves every advantage we can give to it.
As I do watch my training games to select the fittest seafaring AI I realized some other behaviours which didn't change over the games. I Hardly ever saw an Ai (except frisians) build a second carrier producer. So I'd like to ask if the second carrier fits into the preciousness of wares and demand for wares concept of the AI or if something special needs to be considered here. The AI almost never builds a military building just to gain control over more of the mountain and by this misses the necessary gold on crossing the horizon. this means if the last military building is just controlling the space up to the mountain the AI never built even a sentry to own some of the mountains even after gold has been identified as a shortage. I'm not sure whether this could be solved by training.
Last question would how to upgrade the AI. so far I tried 2 possibilities Just adding the best AI to the four files replacing one file or replacing all 4 files with a very good AI and hope for further improvement due to mutation. What would you consider best as results have been of same quality for either strategy.


Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2017-11-19, 23:06

Next observation.
AI tries to build a quarry for example to cut stone which is not reachable due to water. Same with mines or other buildings. from my point of view AI should not try to build on unreachable terrain.


Top Quote
Tibor

Topic Opener
Joined: 2009-03-23, 23:24
Posts: 1377
Ranking
One Elder of Players
Location: Slovakia
Posted at: 2017-11-20, 14:41

hessenfarmer wrote:

As I do watch my training games to select the fittest seafaring AI I realized some other behaviours which didn't change over the games. I Hardly ever saw an Ai (except frisians) build a second carrier producer. So I'd like to ask if the second carrier fits into the preciousness of wares and demand for wares concept of the AI or if something special needs to be considered here.

You mean production sites identified as recruitment, then these are built only one. Necessity of such sites is hard to find out, because carriers are not stocked in warehouse but directly send to a road - or am I wrong?

The AI almost never builds a military building just to gain control over more of the mountain and by this misses the necessary gold on crossing the horizon. this means if the last military building is just controlling the space up to the mountain the AI never built even a sentry to own some of the mountains even after gold has been identified as a shortage. I'm not sure whether this could be solved by training.

Positioning of new training sites if very complex and accessible raw material is one of many inputs. Should be "trainable" but would take a lot of iterations I think...

Last question would how to upgrade the AI. so far I tried 2 possibilities Just adding the best AI to the four files replacing one file or replacing all 4 files with a very good AI and hope for further improvement due to mutation. What would you consider best as results have been of same quality for either strategy.

My opinion is that you should pick 4 winners from about 50 players (so obviously multiple games....). And repeat that multiple times...


Top Quote
Tibor

Topic Opener
Joined: 2009-03-23, 23:24
Posts: 1377
Ranking
One Elder of Players
Location: Slovakia
Posted at: 2017-11-20, 14:44

hessenfarmer wrote:

Next observation.
AI tries to build a quarry for example to cut stone which is not reachable due to water. Same with mines or other buildings. from my point of view AI should not try to build on unreachable terrain.

This is known problem - poor understanding of terrain. AI is able to check if something is reachable from a point but it is very slow, so this is not done. Only dumb check with radius is done.... AI would need some micropartitioinig of terrain and check connectivity for entire microregion. Per single buildable field it is very expensive...


Top Quote
WorldSavior
Avatar
Joined: 2016-10-15, 04:10
Posts: 2091
OS: Linux
Version: Recent tournament version
Ranking
One Elder of Players
Location: Germany
Posted at: 2017-11-20, 14:49

hessenfarmer wrote:

Next observation.
AI tries to build a quarry for example to cut stone which is not reachable due to water. Same with mines

For mines, there is no unreachable terrain. They can always reach everything in their area, no matter which terrain is in between...


Wanted to save the world, then I got widetracked

Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2017-11-20, 20:44

WorldSavior wrote:

For mines, there is no unreachable terrain. They can always reach everything in their area, no matter which terrain is in between...

What I meant was that the Ai tries to build mines in its area which can't be reached by a road due to impassable terrain. Perhaps I shouldn't post lately.


Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2017-11-20, 20:58

Tibor wrote:

hessenfarmer wrote:

You mean production sites identified as recruitment, then these are built only one. Necessity of such sites is hard to find out, because carriers are not stocked in warehouse but directly send to a road - or am I wrong?

No I meant the first building to recruit the second carrier of the tribe (e.g. horses for atlanteans). I never saw the AI build such a building. Of course on crossing the horizon this is not necesdsary but I watched some games on ice wars as well and never saw any of them built which resulted in really weird road networks due to blockage of flags. Just wanted to know whether the alternative carriers are covered by the current necessity algorithm. Don't get me wrong I just want to understand whether it is a matter of training or if we could do something in the algorithm to give the AI the possibility to improve in a certain capability.

The AI almost never builds a military building just to gain control over more of the mountain and by this misses the necessary gold on crossing the horizon. this means if the last military building is just controlling the space up to the mountain the AI never built even a sentry to own some of the mountains even after gold has been identified as a shortage. I'm not sure whether this could be solved by training.

Positioning of new training sites if very complex and accessible raw material is one of many inputs. Should be "trainable" but would take a lot of iterations I think...

It's not about placing training sites. Its about conquering the mountains where the desprately needed metal ores are located. Even the fastest expanding Ai stopped at the border of the mountain, if not "accidentally conquered by a bigger military building. It would have cost them only a sentry to take control over the mountains but they never build one although building lots are constantly available.

My opinion is that you should pick 4 winners from about 50 players (so obviously multiple games....). And repeat that multiple times...

ok will try that.

Regarding the check for unreachable ressources or build spaces perhaps this could be done by scanning the map once for all impassable fields and include them in a table which could be checked before the building is placed, if this would be not so much consuming.


Top Quote