VR Vocabulary Trainer — Master Thesis Repository

1) Project Overview

• Goal: Demonstrate how VR supports learning models via an interactive Vocabulary Trainer that lets learners see, hear, and manipulate 3D word objects.
• Hardware: Meta Quest 2 (PC-VR via Link/Air Link).
• Core idea: Compare learning modes with increasing interactivity:
  • M1 — Seeing & Hearing (baseline; no grabbing)
  • M2 — Seeing, Hearing & Controller interaction
  • M3 — Seeing, Hearing & Hand-tracking interaction

---

2) Key Features

• Three Modes (M1–M3): Toggle interaction depth from passive viewing to full hand-tracking.
• Teleport Locomotion: Predefined hotspots keep users oriented at each learning table (reduces cybersickness).
• Audio Pronunciation: Point to a speech-bubble on each object’s canvas to hear the word.
• Languages: English, German, Vimmi (fictional; ensures zero prior knowledge).
• Basket Mechanic: Drop an object into the basket to “mark as learned” — it despawns after a short delay.
• Tutorial Scene: Teaches locomotion and grabbing with controllers and hand-tracking.

---

3) General Architecture

3.1 Scenes

• StartMenu → pick Mode (M1/M2/M3), Order (fixed/random), Source→Target language (default: EN→VIM).
• Scene1 / Scene2 / Scene3 → neutral rooms with wall tables, teleport points, and object stations.
• Tutorial → basic movement + interaction walkthrough (controllers & hand-tracking).

3.2 Interaction & Locomotion

• Locomotion: Teleport hotspots + viewpoints (keeps the user facing each object).
• Grabbing:
  • M2: Controller grab (Meta XR Grab Interactable)
  • M3: Hand-tracking grab (Meta XR Hand Grab Interactable)
• Object UI: Per-object canvas (title labels; speech-bubble triggers audio). Ray/pointer for controller; pinch/point for hands.

3.3 Data / Assets Layout

Each learnable object lives under Resources/LearnObjects/<objectName>/:

<objectName>.prefab      # 3D model prefab
eng.mp3                  # English audio
ger.mp3                  # German audio
vim.mp3                  # Vimmi audio
loc.txt                  # "German,English,Vimmi" (CSV-style single line)

At runtime, the backend loads these assets, creates LearnObject instances, and spawns canvases and prefabs (random or fixed) at predefined positions.

---

4) Getting Started

1. Install
   • Unity (version compatible with Meta XR Interaction SDK)
   • Meta Quest Link (for PC-VR) and Meta XR All-in-One SDK from the Asset Store
2. Clone & Open
   • git clone https://github.com/CyberNord/MS_VR_SS2024
   • Open in Unity; import Meta XR packages if prompted
3. Target
   • For testing, use PC-VR: connect Quest 2 via Link/Air Link
4. Play
   • Enter Play Mode in StartMenu.unity, choose mode and language, press Start
   • Or Build & Run for PC-VR

---

---

4.1) External Assets

This project requires several external 3D assets that are not included in this repository due to license restrictions.
Please refer to /Docs/ExternalAssets.md for a complete list and download links.

---

5) Usage

• Start Menu Options
  • Mode: M1 (no grab) / M2 (controller) / M3 (hand-tracking)
  • Order: Fixed randomized vs. randomized
  • Source→Target: e.g., EN→VIM, DE→VIM
• In-Scene
  • Teleport to each station
  • Read labels (mother tongue + target)
  • Point speech-bubble to play audio
  • M2/M3: Grab objects; drop into basket when learned

---

6) Research Context & Results

• Built to probe embodied learning and VARK benefits in VR.
• Two expert studies: SUS (good→excellent), NASA-TLX, no cybersickness reported; tutorial scene improved UX.
• No long-term retention test included (future work).

---

7) Acknowledgements

Developed at Johannes Kepler University Linz. For theoretical background, design decisions, and study methodology, see the accompanying Master’s Thesis (Development and Evaluation of a VR Application in a Learning Environment).