Jump to content

Is the AI In Regicide Supposed to Charge Their Hero Out?


Palaiogos
 Share

Recommended Posts

I have come to a conclusion that regicide is the easiest mode. This isn becaus the AI isn't preserving their hero until the very end. It charges it out in that first rush that they usually do 5-10 minutes in the game. When I first played, I lost because their hero was actually pretty good, but then I used the Seleucids, or the Macedonians and they wrecked their hero. The game was over in about 10 minutes. I wonder if you can make the AI more preservative of their heroes. A good example is to make a box formation and keep the hero at the back of that.

  • Like 1
Link to comment
Share on other sites

In terms of the Hannibal AI framework, it is thinkable that you script your own group, e.g. Royal Guard. Its behaviour can be to stalk the hero or keep him/her safe, depending on how you define protection.

The group can easily summon other citizens help defend the hero on losses or during critical situations.

The other AIs can do this too, it requires more coding and especially hard-coding though. Unless we have neural networks or other cool algorithms going already.

Edited by Radagast.
  • Like 1
Link to comment
Share on other sites

Simple regicide support for AI was added in r18731. Currently the AI sends the hero into fight if it has at least 80% HP and retreats it to the next temple / civic center if it reaches 70% and works surprisingly well. Of course if the hero is a seleucid elephant, it can't be garrisoned anywhere and is an easy target.

The AI has no logic to use formations and things like luring AI enemies into towered areas is possible.

 

  • Like 1
Link to comment
Share on other sites

what should he have done? As it was said before, the Hannibal AI could not be finished in the manner it was back then. Contrary to the engine the AI framework is limited if it tries to invent it all within AI context only. This does not work as could be seen in the many workarounds in the Hannibal code.

 

Quote

The AI has no logic to use formations and things like luring AI enemies into towered areas is possible.

Now this is where tactics start. AI attacks from several sides, guerilla groups hidden in enemy / besieged territory. Rescuing the fallen king out of battle like the Spartans did in the Greco Persian Wars.

Edited by Radagast.
  • Like 1
Link to comment
Share on other sites

@Agent_Blair

@agentx actually left because his way of working was different than the one from the team. He wants to come back but I think he wants a more stable framework to developp is AI on. There was trouble with the fact each alpha breaks everything.

Also it there was some issue with Itms at one point because of a miscommunication leading to an argument where I managed to keep things down.

But I guess he is the best one to answer for himself on how hedlike to participate.

On the point Niektb has raised I guess it's easier now that we have wiki pages sorting that out

Edited by stanislas69
Link to comment
Share on other sites

 

1 hour ago, niektb said:

When I spoke him it had nothing to do with the limitations of the AI framework. Instead, it was that he had to adapt the changes in svn. (He found it troublesome to go through dozens of commits to scan for AI-related changes)

Isn't this no contradiction to what I said? Because his code style actually is sensitive to changes. And the outlook was even worse, not only but also due to the isolated invention and reinvention of the wheel.

It may be helpful to study the code in detail. So far there are a dozen forks on Github, all without any activity. The complexity / heavily functional lambda code style seems to stop people (as I often said, code style is absolutely irrelevant when it comes to function, because try to convince overlords that you had to fiddle with code style which is why your code was not ready in time to update the plane/car before it crashed due to this lack of feature).

Finally as long as this complexity is not understood, it is easy to blame other people of misinformation. But do as you like, because for me you are free, people are  free. You can decide if you want to be on slave holder side or our world citizen side.

--

@stanislas69:

It was my opinion that I stated as I studied the code in detail (because I requently updated Hannibal, coded group interaction among others) and warned agentx back then of the upcoming issues. I even wanted to provide pull requests when I felt that he hated me after the 0BC affairs and thus stopped the undertaking.

I do not speak for him, he is a free man. I just spoke out loudly what concerns I saw. Sorry for that. How could I come to state my opinion.

The Itms quarrel already started when I was still active. And it was not just Itms, it was historic/ben too. Among others. Actually agentx had a difficult standing. Much more difficult than me despite my continued confrontations with the management alliances feneur or especially Niek and thamlett (who often talked about what big a modding team they - the great - command and manage despite our no hierarchy arrangement after the Romulus rule) and despite my craziness after epic and saga and my nastiness and agentx' expert code - which is strange.

 

--

Back on topic I just wanted to answer how easy it is to adapt agentx' group engineering to support fancy, dynamic royal protection to counter regicide. I hope he is not angry for me talking out about some of the fruits his steady effort back then produced. :|

Edited by Radagast.
Link to comment
Share on other sites

(Only saw petra code so far and it shouldn't be hard to send a guard command and plot attacks on enemy heroes, finding the people to write the code is. Indeed it is annoying to not be able to access the simulation directly from the AI, but that is due to the thing running in a separate thread, thus having to copy everything, isn't it?)

Link to comment
Share on other sites

I started working on some things I wanted to see in DiplomacyManager and thought it would have been nice to have an "0ad Lite" version in which case the engine / basic mod (1 map, 2 civs) / and the AI are decoupled for this type of testing.  I could see what the frustration would be working on AI but in general Petra seems to be (guessing) a little unmanageable when there are multiple objectives beyond conquest.  Regicide from what I saw, would have to give nearly every manager class it's own behavior for just that type of game play.  In other words a Petra forked specifically for Regicide, etc.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...