Capstone 1: AI-Driven Video Game Testing (Forest Escape)
Forest Escape is an AI-powered game testing framework built to automate quality assurance in video game development. Its primary objective is to utilize artificial intelligence to replicate realistic player behaviour and conduct comprehensive game testing without human intervention. By employing a reinforcement learning agent within a 3D forest-themed level, the system enhances the speed, precision, and scalability of testing. This addresses the challenge of slow, labour-intensive manual QA by reducing human effort in bug detection and gameplay balancing, while enhancing overall game quality. The framework functions as a proof of concept for AI-driven virtual game testers—capable of navigating environments, identifying issues, and validating whether the game aligns with design standards
Capstone 2: GhostRoute Framework (Smart Navigation Unity Toolkit)
GhostRoute is an AI-powered QA toolkit for 3D game development that uses Unity ML-Agents and PPO to automate bug and design flaw detection. Testing showed a 100% success rate on known maps, with reliable navigation but limited generalization to unseen areas. Visual analyses confirmed strong central exploration but weaker edge coverage. While currently Unity-focused and lacking advanced features, GhostRoute offers small studios a scalable, cost-saving alternative to manual QA, with future improvements needed in transfer learning and adaptability.
Capstone 1: AI Video Game Testing
Capstone 2: The GhostRoute Framework
Capstone Reflection: The GhostRoute Framework
