Leveraging Unreal Engine 5 to Develop Synthetic Data Tools for Lunar Surface AI enabled Robotic Applications

This project aims to research and develop a tool using powerful game engine (Unreal Engine 5) and digital elevation models to produce large amounts of photo-realistic training data with intuitive interfaces for end-users. The project further aims to research methods of verification of the data generated, analyse its Sim-to-Real predictivity and test the usability of the data generated on some machine learning based models popularly used in robotic perception applications (i.e. depth perception. hazard detection). A key goal for this project is the development of a framework for generating synthetic Lunar surface imagery that is easily expandable and upgradable through future research and toolset integration.