[General bot combat questions] Lets talk about a combat system...

[Därkness]

Background:
For the past week or so my project has been behavior trees, and with the help of aigamedev, apoc's treesharp, and a bunch of articles by game developers I managed to complete the bare-bones of a simple BT in delphi.

So its testing time . Long-term goal: make a PvP bot with a great AI/pathfinding/combat. I was looking at creating a "combat sub-tree", which has branches account for different situations like moving, stunned, not moving, and target is at x distance etc, each branch with a priority based selector, with children loaded from a file (different abilities key presses etc. Each will have decorators for conditions).

tl;dr: I'm trying to make a bot to trial a combat system, to send a keypress to use an ability at the right moment (not on cooldown, can use it, best for situation etc).

So my question:
How do you guys implement a more "complex" combat system?

Also what are the different branches i should have?

[Come Undone]

I would have the main branch based on movement:

Let's call the first one, running after the health- and enemystate checks; canmove.
If it's possible to move, we have no problems at the moment and we can hit them - carry on to next state check; hasbuffs, or perhaps -had-buffs.

Reading from former buffs(debuffs) you had which made your character not passing the canmove check, which would generally have cooldowns if cast by enemy players (yeah, it's pvp right?). Making a file with enemy stuns, hamstrings cooldowns could make it possible for you to calculate how much time your automa controlled character have to perform actions until next stun. And so on, it will never be reliable though, since it's PvP and many players can interact in combat. So it's something really tough to perfect... And very complex.

Something you WOULD want, and with ease could have is: a check on how many opponents are near eachother, and if your automa would stand a chance, or rather prioritize finding other opponents.

Good luck, and try not to make something too complex, simplicity often wins.