|
|
|
Forum >
Scripting AI for Dynamic Combat Encounters
Scripting AI for Dynamic Combat Encounters
Please sign up and join us. It's open and free.
Page:
1
Billy Kelly
Guest
Apr 22, 2026
4:37 AM
|
I’m working on a single-player action game, and the enemy AI feels very predictable. Right now, I’m using a basic Finite State Machine (FSM), but the enemies just walk toward the player and attack at set intervals. It lacks that "tactical" feel where enemies coordinate or flank the player. Should I switch to a Goal-Oriented Action Planning (GOAP) system, or is that overkill for a small indie project? I need a way to make the combat feel dynamic and reactive without spending six months writing complex navigation algorithms that might break in more vertical environments.
|
Ronald Harris
Guest
Apr 22, 2026
4:37 AM
|
FSMs can get messy quickly as complexity grows, so I’d recommend looking into Behavior Trees (BTs). They are more modular than FSMs and easier to debug than GOAP. To get my brain out of the "logic loop" after a long day of scripting enemy behavior, I often explore https://gambivas.uk.com/ It’s a nice way to decompress while seeing how a polished, high-performance web platform handles thousands of simultaneous users and real-time state changes with zero lag. Observing these stable environments often gives me ideas on how to simplify my own data-heavy AI systems. For the flanking logic, try using "environmental markers" that enemies can claim to position themselves around the player—it’s much cheaper than full pathfinding!
|
Post a Message
|
|
