Table of Contents

Introduction
Built with
Status
Features
How to Run
    In Unity Editor
    From Build (Standalone)
Sample Results
    Scene 1: Sign in
    Scene 2: Lobby List
    Scene 3: Lobby Room
    Scene 4: Play Scene
    Map Overview and Zone Partitioning
    InfoPoints
Links
Current Limitations and Future Development
    Current Limitations
    Future Development
Credits

Introduction

This project is a graduation thesis presents the development of a server-authoritative 3D multiplayer FPS game built using Unity and Unity Gaming Services. The system utilizes Unity Relay for serverless connectivity, Netcode for GameObjects (NGO) for real-time synchronization of networked entities, and Unity Lobby for matchmaking and session management.

Beyond basic multiplayer functionality, the project focuses on the integration of AI-controlled bots operating in a networked environment. These bots are fully controlled by the host and synchronized across all clients, ensuring consistent behavior and fairness in gameplay. A hybrid Finite State Machine – Behavior Tree (FSM–BT) architecture is employed to manage bot decision-making, balancing structured state control with flexible behavior execution.

The goal of this thesis is to design, implement, and evaluate a scalable multiplayer FPS architecture that supports both human players and network-synchronized AI agents using modern Unity networking services.

Built with

• 
• 
• 
• NGO (Netcode for GameObjects)
• Unity Relay
• Unity Lobby
• Behavior Designer

Status

Completed and closed after successful Graduation Thesis defense. Only minor refinements and bug fixes remain.

Features

• Map game: the Italy map inspired by Counter-Strike
• Character model: fully humanoid player model
• Full basic movement sync (idle, walk, run, jump)
• Complete weapon system including:
  • Weapon: Rifle, Sniper, Pistol, Melee, and Grenade
  • Core mechanics: shooting, aiming, reloading, and weapon switching
  • Server-authoritative damage system with hit and death effects to prevent cheating
• Fully playable match loop:
  • Scoreboard system
  • Win/Loss (Victory / Defeat) conditions
• AI Bots with a hybrid FSM–Behavior Tree architecture:
  • High-level states: Idle, Patrol, Combat
  • Detailed behaviors implemented as BT tasks: LookAround, ScanArea, Seek, AimAtPlayer, Attack
  • Hierarchical pathfinding system combining graph-based and navigation mesh approaches:
    • Dijkstra’s algorithm is used to compute paths between zones at the strategic level
    • Unity NavMesh is used for local pathfinding and movement within individual zones
• Zone-based spatial reasoning system:
  • ZoneData and InfoPoint structures support area scanning, tactical positioning, and target pursuit
  • These spatial data are baked in the Unity Editor and serialized as ScriptableObjects
  • At runtime, the system loads and use these data to support AI decision-making and navigation processes
  • This approach enables bots to evaluate visible areas, select navigation targets, and perform contextual pathfinding

How to Run

In Unity Editor

1. Open the project in Unity (version 2022.3+ recommended)
2. Install all dependencies via Unity Package Manager
3. Press Play to start

From Build (Standalone)

1. Ensure the project is linked to Unity Services (Relay + Authentication enabled)
2. Build the project via File > Build Settings
3. Launch the game executable

Note: Internet connection is required for Unity Relay to work properly.

Sample results

Scene 1: Sign in

Scene 2: Lobby List

Create lobby with info:

Scene 3: Lobby Room

Scene 4: Play Scene

Map Overview and Zone Partitioning

Top-down view of the Italy-inspired map illustrating zone partitioning. The map is divided into multiple zones, each managing its own spatial data, forming the foundation for graph-based pathfinding algorithms Dijkstra.

InfoPoints

Visualization of InfoPoints in the T_Spawn area of the map. Yellow markers - InfoPoints, blue markers - TacticalPoints, and purple markers - PortalPoints. AI bots utilize the spatial data provided by these point types to perform area scanning within the zone.

Links

• Video demo: Link video youtube
• Flow chart: Link draw.io

Current Limitations and Future Development

Current Limitations

• UI/UX is still basic
• Player character animations are not natural; movement lacks polish and proper animation logic
• The game still experiences noticeable lag
• High latency difference between host and clients causes gameplay imbalance
• AI bot movement is sometimes unnatural, particularly when navigating sharp corners or narrow passages
• Area scanning and perception behaviors remain mechanical and lack human-like variation

Future Development

• Improve UI/UX (main menu, HUD, scoreboard, lobby navigation)
• Add game modes (Team Deathmatch, Bomb Defusal)
• Polish character animation, effects, and network optimization
• Add save/load, progression, or stats system (optional for polish)
• Develop more advanced combat behaviors (seeking cover, retreating)
• Enable bots to use other weapon types(Grenade, Melee, ...)
• Enhance perception systems (sound-based target detection)
• Increase the number of AI bots, group behaviour

Credits

Developed by Kieeran and Haiseus
Special thanks to:

• Haiseus - for your help with this project

And thanks to everyone who have helped with suggestions and feedback!