Roleplaying Open World Prototype

Gameplay Systems Overview

This project is a gameplay systems-focused open world prototype inspired by modern crime sandbox games. The primary focus of the project was designing scalable AI systems that could seamlessly transition between on-foot and vehicle-based pursuit while maintaining responsive gameplay behavior.

The project emphasizes:

  • AI pursuit systems

  • Vehicle chase behavior

  • Multi-state AI decision making

  • Vehicle possession transitions

  • Obstacle-aware driving logic

  • Gameplay system architecture

  • Debugging and system visualization tools

⭐Featured Gameplay Systems⭐

Police Pursuit AI System

System Summary

The police pursuit system was designed to create dynamic chases where AI officers can pursue the player both on foot and in vehicles, depending on distance, context, and vehicle availability.

System Support

  • On-foot pursuit

  • Vehicle pursuit transitions

  • AI vehicle entry and exit

  • The possession handoff between the officer and the vehicle

  • Obstacle-aware vehicle driving

  • Look-ahead steering behavior

  • State-driven pursuit logic

AI Detects Player

→ Begins On-Foot Chase

→ Evaluates Distance To Player

→ Searches For Available Vehicle

→ Requests Vehicle Entry

→ Transfers Possession To Vehicle

→ Begins Vehicle Pursuit

→ Adjusts Steering Toward Dynamic Aim Point

→ Avoids Obstacles During Chase

→ Returns To On-Foot Pursuit When Necessary

Technical Breakdown

The AI Combat Coordinator acts as a central decision manager for enemies in combat.

When an enemy detects the player and reaches a suspicion threshold, it registers itself with the coordinator. The coordinator evaluates the list of active enemies and grants attack permission to one AI at a time.

Other enemies remain active but are placed in a waiting state, where they circle the player and maintain pressure until the coordinator grants them permission to attack.

This approach prevents enemy overlap while maintaining the illusion of coordinated group combat.

Blueprint Logic

Enemy Registration

When an AI's suspicion meter exceeds a threshold:

  1. AI sends a RequestAttackPermission call to the Combat Coordinator.

  2. The coordinator stores the AI reference in an Active Enemy Array.

  3. If no attacker is currently active, the coordinator selects one.

Attack Selection

The coordinator:

  • Checks if CurrentAttacker is empty

  • Selects an AI from the registered enemies

  • Sets that AI as CurrentAttacker

  • Updates its CanAttack flag

This flag is read by the AI’s Behavior Tree to enter the attack state.

Behavior Tree Integration

AI behavior trees reference the coordinator through blackboard variables.

Typical flow:

Selector
 ├── Attack Player (if HasAttackPermission)
 ├── Circle Player (if WaitingForTurn)
 └── Investigate / Chase Player

The HasAttackPermission value is controlled by the coordinator.

Circling System

Enemies waiting for their attack turn move around the player instead of standing still.

The circling position is calculated by:

  • Getting the player location

  • Generating an offset position around the player

  • Using vector rotation to distribute enemies around the player

This ensures enemies maintain spacing while remaining threatening.

Challenges & Solutions

Multiple AI Attacking at Once

Initially, enemies would attack simultaneously when they detected the player.

Solution

Implemented a centralized attack permission system that ensures only one AI can enter the attack state at a time.

Idle AI Standing Still

Enemies waiting for their turn felt unnatural when standing still.

Solution

Added a circling behavior that generates dynamic positions around the player, allowing enemies to reposition while waiting.

Combat Flow

Without coordination, combat encounters felt chaotic and overwhelming.

Solution

The coordinator system regulates attack timing and enemy positioning, creating combat that feels intentional and readable for the player.