[Master Thesis / Guided Research] Live Streaming of Submap Gaussian Splatting to VR: Comparing Region-of-Interest Annotation Strategies for Waypoint Guidance
01.09.2026, Abschlussarbeiten, Bachelor- und Masterarbeiten
This thesis focuses on bringing a submap-based 3D Gaussian Splatting (3DGS) reconstruction into an interactive VR environment through live streaming to Unity. The project will extend an existing 3DGS pipeline and Unity rendering plugin to support progressive scene loading as users move through a large reconstructed environment.
The main research focus is the design and comparison of different Region-of-Interest (ROI) annotation strategies in VR. Users will be able to mark interesting locations using interaction techniques such as controller-based ray casting, direct hand/gesture interaction, or gaze-based selection. Each marked location will be stored as a persistent waypoint and accompanied by a visual or navigational guidance cue.
A comparative user study will evaluate the different annotation techniques based on factors such as accuracy, task completion time, error rate, workload, usability, and user preference. The results will be used to identify suitable interaction techniques and derive design recommendations for annotation and wayfinding in streamed 3DGS environments.
The thesis combines computer graphics, real-time 3D streaming, Unity/C#, VR interaction, and user-study methodology, providing both an implementation component and an empirical research component.
Your skillset
* Program confidently in C/C++/C# / or expertise in Unity
* Understand the basic concepts of **3D Gaussian Splatting (3DGS)** and real-time rendering
What you should do during your thesis:
* Implement **live/progressive streaming of 3D scene data** from our backend to a VR application
* Develop for **VR/XR**, including interaction with an HMD and controllers/hands
* Design and implement different **VR interaction techniques** for selecting and annotating regions of interest
* Store and manage **persistent waypoints/annotations** in a 3D environment
* Implement basic **wayfinding and guidance visualizations** in VR
* Design controlled **user-study tasks and experiments**
* Collect and analyze **quantitative and qualitative user-study data**
Kontakt: Hannah.schieber@tum.de, hex-thesis.ortho@mh.tum.de


