Comprehensive research and technology evaluation for the Portalt project
We evaluated multiple technical approaches to address the core challenge of enabling global access to IBM's Innovation Centre without physical travel, while maintaining interactivity and scalability for enterprise and educational use cases.
Baseline Solution
Implementation: Static panoramas with clickable hotspots (e.g., IBM Hursley Museum virtual tour).
ENGAGE XR, Virbela
Implementation: Commercially available enterprise VR platforms with established feature sets.
Selected Approach
Implementation: Unity-based development with Ubiq's networking framework, IBM Watson Assistant, and Granite AI for contextual queries.
Our research evaluated various devices for both administrative interfaces and immersive experiences to determine the optimal hardware approach for Portalt.
Device Category | Selected Devices | Key Benefits | Considerations |
---|---|---|---|
Admin Interface Web-Based Built with Next.js & Electron.js |
Desktop Windows, macOS, Linux |
|
Primary development target with complete functionality |
Mobile/Tablet iPadOS, Android |
|
Limited 3D preview capabilities and reduced features |
|
VR Experience Immersive Unity-based application |
Meta Quest Series [1,2,3] Quest 3, 3S, and 2 |
|
Primary target platform with tiered optimization for different models |
Apple Vision Pro [6] Premium MR device |
|
Secondary target for premium enterprise settings |
|
Future Devices [8,9] AR Glasses, Mobile AR |
|
Architecture designed to accommodate future devices |
Research findings indicate that Meta Quest devices offer the optimal balance of accessibility (standalone operation), technical capabilities (hand tracking, passthrough), and market penetration (80% share) [4,5]. The web-based admin interface ensures cross-platform accessibility while providing rich functionality when used with Electron on desktop systems.
Asset Management, User Authentication, Environment Configuration
Our research identified the optimal technology stack for developing Portalt's multi-layered architecture, focusing on compatibility with VR hardware, integration capabilities, and performance requirements.
Based on our requirements analysis, we determined a four-layer architecture as the most effective approach for Portalt:
Layer | Description | Components |
---|---|---|
Frontend Layer | User interface components for different user types |
|
Authentication Layer | User identity and access control |
|
Storage Layer | Data persistence and file storage |
|
Compute Layer | Real-time interactions and AI processing |
|
Component | Selected Technology | Key Benefits | Alternatives Considered |
---|---|---|---|
VR Engine |
Unity C# programming language |
|
|
Networking |
Ubiq Open-source framework |
|
|
Admin Platform |
Next.js +
Electron.js TypeScript |
|
|
UI Components |
Radix UI +
Tailwind CSS For admin interface |
|
|
Our research into multiplayer VR frameworks led us to select Ubiq, an open-source networking framework specifically designed for social VR experiences. Ubiq provides several critical capabilities for Portalt's implementation:
Ubiq Component | Portalt Implementation |
---|---|
NetworkScene | Acts as the central mediator for all network events in Portalt activities. Manages the synchronization of presenter movements, interactive elements, and shared 3D models across all participants in a session. |
NetworkId System | Provides unique identifiers for all networked objects in Portalt scenes, ensuring that interactions with quantum computing models, healthcare simulations, and educational content are correctly propagated. |
RoomClient | Handles the organization-based access control in Portalt, connecting activity sessions to specific IBM organizational spaces and managing user permissions within those environments. |
Genie RAG Service | Implements the contextual AI assistant that can answer questions about IBM products, systems, and educational content by retrieving information from associated documents and using Granite LLM for response generation. |
Voice Processing | Utilizes Whisper API for converting participant questions into text, then routes the processed text to the appropriate AI service (RAG or direct response) based on content and context. |
Our research indicated that Ubiq's deterministic networking model is particularly well-suited for IBM's demonstration needs, where technical accuracy in interactive models (e.g., quantum computing visualizations) is critical. The Ubiq Genie extension enables Portalt to integrate voice-activated AI assistance that maintains awareness of the demonstration context, providing relevant information about specific products or concepts being showcased.
Component | Selected Technology | Key Benefits | Alternatives Considered |
---|---|---|---|
Authentication Service |
Clerk User management system |
|
|
Pairing System |
Custom 6-character codes |
|
|
Token Management |
JWT JSON Web Tokens |
|
|
Component | Selected Technology | Key Benefits | Alternatives Considered |
---|---|---|---|
Local Database |
SQLite with MongoDB-like interface |
|
|
Cloud Storage |
IBM Cloud Object Storage |
|
|
Asset Format |
glTF/GLB 3D asset format |
|
|
Component | Selected Technology | Key Benefits | Alternatives Considered |
---|---|---|---|
Multiplayer Server |
Ubiq Server NodeJS implementation |
|
|
AI Framework |
Genie Framework Modular service architecture |
|
|
Speech Processing |
Whisper API For speech recognition |
|
|
LLM Service |
Granite LLM 3.2B parameter model |
|
|
Our database design research focused on creating a schema that supports the complex relationships between entities while maintaining flexibility for future extensions:
Collection | Purpose | Key Fields |
---|---|---|
Activities | Store activity information |
|
Documents | Store document metadata |
|
ActivityDocuments | Link activities to documents |
|
Organizations | Store organization data |
|
Scenes | Store scene information |
|
PairingCodes | Store organization access codes |
|
Our research led us to implement a hybrid approach combining SQLite's ACID compliance with document-based storage flexibility. This allows for both structured relationships and dynamic data storage, with universal fields (id, data, timestamps) and JSON document storage in the data field.
Based on comprehensive research, we've designed Portalt with a four-layer architecture (Frontend, Authentication, Storage, and Compute) optimized for immersive VR experiences with enterprise-grade administration capabilities. Our core technology stack combines Unity and Ubiq for VR development with Next.js and Electron.js for the cross-platform admin interface. This approach balances immersive quality with accessibility, targeting Meta Quest devices as primary hardware while maintaining flexibility for future AR integrations.
IBM technologies form the backbone of our cloud infrastructure, with IBM Cloud Object Storage providing enterprise-grade asset management, IBM Watson offering advanced AI capabilities, and Granite LLM (3.2B parameter model) enabling context-aware conversations with low latency. This IBM-centered approach allows seamless integration with IBM's existing systems while providing the robust infrastructure needed for global scalability. The architecture specifically optimizes for IBM's use cases—quantum computing demonstrations, healthcare simulations, and educational scenarios—through deterministic networking and RAG-based contextual information retrieval.
Security and data management needs are addressed through multiple integrated solutions: Clerk for enterprise-focused authentication with organization management, a custom pairing system for simplified VR access, JWT for secure token management, and a hybrid SQLite approach combining relational database reliability with document-based flexibility. This comprehensive technology stack provides IBM with a customized solution that surpasses generic enterprise VR platforms through specialized AI integration and tailored demonstration workflows.