If you have trouble registering or do not receive a verification e-mail, please ask for help in chat.

adaptative AI

Request new features or present your ideas.
User avatar
freem
Marauder
Posts: 171
Joined: Mon Feb 11, 2013 1:02 pm UTC

adaptative AI

Postby freem » Wed Sep 15, 2021 10:39 pm UTC

Hello.

What do you think about adaptative AIs?
That is, AI which balance imbalanced teams. Teams can be imbalanced by 2 things: the number of players, and the skill of players. How would you imagine having AI to balance those situations while not making infinite games?

This is a complex topic, so please, keep in mind that every input is welcome, as it will help shaping a better idea of how that could be implemented.
User avatar
Gireen
Graphic Designer
Posts: 266
Joined: Wed Mar 07, 2012 1:26 pm UTC
Location: Germany
Contact:

Re: adaptative AI

Postby Gireen » Thu Sep 16, 2021 10:09 pm UTC

Both teams don't need to have exactly the same skill.
But the difference should not be to big.
fear ma engrish :granger:
User avatar
freem
Marauder
Posts: 171
Joined: Mon Feb 11, 2013 1:02 pm UTC

Re: adaptative AI

Postby freem » Sat Sep 18, 2021 2:23 pm UTC

The question is more about having AI's skill changing as people join or leave, depending on people's skills, not about having same skill level for aliens and humans (that's an interesting question too, though, considering the fact there's few cons to this).
For that, it's needed to evaluate people's skills. The kill/death ratio is a metric, for sure, but it does not apply to building. The score is a metric too, but personally I have no idea how it growth or shrinks, so I'm not fond of using it.
Credit income per minute would not work, because camping is rewarded by 200 credits every 120s spent alive, and attacking buildings does not generate income.

I think one of the problems is: how to evaluate mechanically if something was built in a nice place or not? By checking if it's still alive after some time? How much time, then? How much "score" should this give?
User avatar
Gireen
Graphic Designer
Posts: 266
Joined: Wed Mar 07, 2012 1:26 pm UTC
Location: Germany
Contact:

Re: adaptative AI

Postby Gireen » Sat Sep 25, 2021 12:35 pm UTC

Spend BP in relation to lost BP
If they build something that isn't destroyed its good :grin:

But what need do we have to get builder skills? They don't participate in combat.

The combined skill of a team changes all the time, as you said by joining and leaving players but also the individuals skill can also change over time.
Any rough estimate that changes all the time slightly is already a good start.

Kills / Death / Assists are still the best measurement i think. It also gives information about how involved a player is in the game.
fear ma engrish :granger:
User avatar
freem
Marauder
Posts: 171
Joined: Mon Feb 11, 2013 1:02 pm UTC

Re: adaptative AI

Postby freem » Sun Sep 26, 2021 12:21 am UTC

so a builder is to be considered a useless one?
User avatar
Gireen
Graphic Designer
Posts: 266
Joined: Wed Mar 07, 2012 1:26 pm UTC
Location: Germany
Contact:

Re: adaptative AI

Postby Gireen » Fri Oct 01, 2021 5:24 pm UTC

That's your conclusion. :tongue:

Let me ask, how should a bot react differently to a good builder compared to a bad one?
fear ma engrish :granger:
User avatar
freem
Marauder
Posts: 171
Joined: Mon Feb 11, 2013 1:02 pm UTC

Re: adaptative AI

Postby freem » Sun Oct 03, 2021 8:32 pm UTC

Gireen wrote:That's your conclusion. :tongue:

Let me ask, how should a bot react differently to a good builder compared to a bad one?


I don't really want bots to adapt their targeting based on the target's skill, I want them to adapt their global skill level so that the team's global skill levels tend to be close. The idea is really about balancing teams, not to victimise a player.

That is, if there's a newbie fighting a pro, the newbie will have SK8 bots and the pro will have SK2 ones. Taken as extreme for the image.

Return to “Ideas & Suggestions”

Who is online

Users browsing this forum: Bing [Bot] and 2 guests