Arena Champions

Take control of a Ludus, conquer reputation at the stores to get the best rewards, buy gladiators and train them well enough so that they will bring the glory back to your Ludus as the new Arena Champions! 

Technical Overview:

Project: Advanced Unity 2D Strategy & Resource Management Game
University: UNISINOS, Brazil
Special Feature: Implementation of Reinforcement Learning
Architecture: ECS with Service Locator Pattern for Scalability

Description:
Developed as a capstone project for a Digital Games degree at UNISINOS, Brazil, this Unity 2D strategy game showcases the cutting-edge application of Machine Learning within the gaming industry. The game leverages reinforcement learning to dynamically balance gameplay, providing each gladiator with a 'learning brain' that evolves based on combat experience. This innovative feature enhances player engagement by progressively intensifying the game's challenge as players advance.

Technical Highlights:

• Machine Learning Integration: Uses reinforcement tree learning (from the book 'AI for Games', by Milington) to autonomously adjust game balance, ensuring a continuously engaging experience for players.
• ECS Architecture: Employs Entity Component System (ECS) architecture, enhancing game performance and maintainability.
• Service Locator Pattern: Implements the Service Locator design pattern, facilitating the integration of expansive systems without complicating the existing architecture.
• Autonomous Agents: Each gladiator is equipped with a neural network, trained through gameplay to improve tactics and decision-making continuously.

Explore the detailed Game Design Document (GDD) for a deeper insight into the game’s mechanics and innovative features: GDD Document