MED5: Immersive Experiences

Immersive Installations with Interaction tasks

Design and evaluate virtual, augmented, or extended reality applications that support subtle, embodied, task-defined interactions.

In this experimental setup, the white bar in the center of the screen helped participants perceive the correct distance 𝑆. They needed to match the size of this bar with the lead car’s bumper while driving. Here, it is wider than the lead car’s bumper indicating that the participant needs to accelerate. After [1].

MED5 Framing:

Multimodal Feedback In Virtual Reality Task: Evaluating the effect of visual, auditory and haptic feedback on task performance, cognitive load and immersion

  • Target users / audience: Who is the end-user or organisational persona this installation serves? (e.g. driving-assistance operators, museum visitors, clinical trainees)
  • Spatial Computing competence: Real-time 3D pipeline across XR/mixed-reality — diegetic UI/UX, multi-modal interaction (gaze, pinch, voice), dynamic environment workflows.
  • Quantitative validation: Instrument the prototype with real-time telemetry (task completion time, error rate, or biometric/eye-tracking data where relevant). (Optional: Compare results to a theoretical model.
  • Performance budget: State target framerate/platform constraints and how they were profiled and met.
  • Job-market link: Name the industry roles/sectors this maps to (automotive HMI, XR training, spatial simulation) so you can articulate it on a CV/portfolio.

References

[1] Vertegaal, Roel, Timothy Merritt, Saul Greenberg, Aneesh P Tarun, Zhen Li, and Zafeirios Fountas. 2025. “Interactive Inference: A Neuromorphic Theory of Human-Computer Interaction.” arXiv. doi:10.48550/arxiv.2502.05935.

Embodied Avatars

Design and evaluate one or more interactive conversational agents with our without LLM backend.

Related work: FirstImpress, Bargum & Hansen, ISAAR paper. Embodied Avatars

MED5 framing

  • Agentic AI integration: Specify whether the agent uses a local/cloud LLM at runtime and how its behavior, latency, and orchestration are controlled — not just “prompted,” but engineered.
  • Architecture & version control: Document the system architecture (modular services, API boundaries) and repo/build setup (Git, CI) so the project reads as team-integration-ready.
  • Ethics & data integrity: Address conversational data handling, privacy, and bias considerations for the LLM backend (per EU AI Act).
  • Stakeholder framing: Identify who would deploy this agent (a company, museum, clinical service) and the value proposition for them.

See the full listing of student theses (2013–2026).