Project Pastir – Prototype Build

🌿 Project Pastir

You are not the Pastir(Shepherd).

You are the one who guides them.

In Project Pastir, you take the role of an unseen force that leads a shepherd and their flock across a harsh, fading world. Wherever you move, life returns—grass grows, and hope follows. But where you leave, the land withers.

Your duty is simple: help them survive.

🧭 Core Concept

This is a survival-focused RTS where your goal is not just to fight—but to endure.

The sheep are not just something you protect—they are the center of your survival.

If they fall, the journey ends.

Think of them as living, moving objectives that must be defended at all costs.

⚔️ Gameplay and Controls Overview

Real-Time Unit Control

Select and command units like a traditional RTS[Pastir the Shepherd and the dog]

→ Click to select

→ Target enemies to attack

Survival Loop (25-Day Prototype)

The game progresses through escalating days

→ Each day increases pressure and difficulty

→ Survive as long as possible within the cycle

Protect the Flock

Sheep act as mobile core units

→ If they die, the game ends

→ Your entire strategy revolves around keeping them alive

Environmental Influence (The “Pastir”)

You shape the land itself

→ Move across the map to bless terrain

→ Grass grows where you go

→ The world decays where you leave

Map Awareness Tools

→ Edge scrolling to navigate

→ Minimap to track threats from afar

→ Card Dial System to jump between units instantly

🧪 Prototype Scope

 This prototype focuses on demonstrating and comparing two different AI approaches within the same gameplay environment:

- Finite State Machine (FSM) — rule-based, deterministic behavior

- Machine Learning Agents (ML-Agents) — adaptive, learning-based behavior using Unity ML-Agents Toolkit

Both systems are implemented to control in-game behavior under identical conditions, allowing direct comparison in terms of:

- Responsiveness
- Adaptability
- Survival performance
- Player interaction impact

The current version includes:

- A 25-day survival cycle as a controlled test duratio
- Core RTS mechanics for consistent player input
- AI-driven enemies and interactions using both approaches

🧠 Research & Testing Phase (Updated)

⚠️ This project is part of an undergraduate research study

This game is being used to evaluate how different AI systems affect gameplay experience and difficulty balancing in a survival RTS environment.

🎯 Research Focus

- Comparing FSM (rule-based AI) vs ML-based agents
- Observing how each approach influences:
     - Game difficulty
     - Player survival outcomes
     - Behavioral adaptability
- Applying machine learning to explore dynamic difficulty in survival gameplay

📊 What’s Being Measured

- How long players survive
- How effectively players protect the flock
- Differences in challenge between AI systems
- Player feedback and perceived difficulty

📌 What this means for you

- You are playing a test build designed for data collection
- Gameplay may vary depending on which AI system is active
- Your experience directly contributes to the research results

🎁 Rewards & Events

🏆 Hall of Fame

Reach the highest sheep count and your name will be added to the game’s future Hall of Fame.

I don't hava the preview yet ~

🎬 Art & Video Contest

Create anything about the game (art, clips, memes, edits)

Total Prize: 500,000 IDR

Winners: 10 people

The banner is also havent done yet

📝 How to Join

Play the game

Prepare your content

Submit via the survey form

Survey and submissiom hasn't open yet!

⚠️ Note

Contest runs during the 1–2 month testing period. Winners announced after it ends.