Core Architecture

Documentation Unreal Engine AI Architecture

Understanding the StateTree + Behavior Tree hybrid architecture used in professional action games.

The StateTree + Behavior Tree Hybrid

This plugin implements the AAA-standard AI architecture used in professional action games. Understanding this architecture is key to using the plugin effectively.

Why This Architecture?

The Problem with Traditional Approaches:

Most developers build AI using a single massive Behavior Tree. This leads to:

❌ Unmaintainable trees with hundreds of nodes
❌ Mixed concerns (strategy + execution in one place)
❌ Difficult debugging and iteration
❌ Poor scalability as complexity grows

The Professional Solution:

Separate strategic thinking from tactical execution:

StateTree (The Brain) - High-level decision making:

┌─────────────────────────────────────┐
│ StateTree: Strategic Layer          │
├─────────────────────────────────────┤
│ • Evaluates combat context          │
│ • Scores available actions          │
│ • Manages state transitions         │
│ • Controls movement strategy        │
│ • Handles target acquisition        │
└─────────────────────────────────────┘
          ↓
┌─────────────────────────────────────┐
│ Action Execution: Tactical Layer    │
├─────────────────────────────────────┤
│ Option A: Gameplay Ability          │
│   → Simple: Play montage, apply dmg │
│                                     │
│ Option B: Behavior Tree Sequence    │
│   → Complex: Multi-step, branching  │
└─────────────────────────────────────┘

Benefits:

✅ Clear Separation - Strategy decisions separate from action execution

✅ Scalability - Add new actions without touching state machine logic

✅ Debuggability - Isolate issues to strategy or execution layer

✅ Maintainability - Each layer has single responsibility

✅ Industry Standard - Same pattern used in major action games

Pro Tip: When to Use Each Execution Mode Use Gameplay Abilities for simple attacks (play montage → damage). Use Behavior Tree sequences for complex multi-step actions with conditional branches (charge attack → check stamina → branch to heavy or light version).

System Flow

StateTree polls ActionEvaluationComponent each tick
Component gathers context: distance, angle, health, threat, tags
ActionSet scores all available actions using:
- Distance and angle evaluations
- Health-based modifiers
- Cooldown status
- Context tags (player blocking, low health, etc.)
- Novelty decay (anti-repetition)
Best action selected and executed as:
- Gameplay Ability (via AbilitySystemComponent), OR
- Behavior Tree sequence (via temporary BT component)
Movement system evaluates 16 directions simultaneously:
- Scores each for distance, angle, avoidance
- Blends pathfinding with direct input
- Applies strafing, retreat, or approach behavior

✅ Production-Ready Enemy AI - Intelligent tactical combat behaviors
✅ Modular Component System - Mix and match for different enemy types
✅ Data-Driven Configuration - Designers tune without code
✅ Blueprint-Friendly - Full Blueprint support, C++ optional
✅ Demo-Included - Working enemy to study and extend
✅ Professionally Architected - AAA-standard StateTree/BT hybrid
✅ Scalable - Works for simple enemies and complex bosses
✅ Keeps Improving - I develop games with the systems I have made. So this plugin will keep improving!

What It's NOT (Yet)

❌ NOT a Complete Game Template - Combat systems only, not a full game
❌ NOT Player Character Combat - Focused on enemy AI (player template coming Q1 2025)
❌ NOT Multiplayer-Ready - Single-player focused (replication planned Q2 2025)

Scope & Expectations

This plugin provides:

Enemy AI decision-making and tactical movement
Melee combat hit detection
Action scoring and execution framework
Multi-enemy coordination
Component-based architecture

You still need in order to make a game:

UI/HUD (health bars, lock-on indicators, damage numbers)
Level design and encounter placement or spawning the enemies
Progression and difficulty balancing