Simulation-Based Training Platform

Part of the worlds-biggest-software-project initiative.

An AI-native, open-source platform for physics-based immersive training simulations across medical, industrial safety, and defence domains.

Simulation-Based Training Platform provides physics-accurate virtual environments where learners practise high-stakes procedures -- surgical operations, industrial safety drills, equipment maintenance, emergency response -- without risk to people or property. It targets training organisations, hospitals, military units, and industrial operators who need repeatable, measurable skill development that traditional classroom instruction, mannequins, and live exercises cannot deliver affordably or safely.

Why Simulation-Based Training Platform?

Every major platform is a closed garden. Osso VR, FundamentalVR, Oxford Medical Simulation, and Strivr all use proprietary scenario formats with no open API for programmatic content creation or result ingestion. An open scenario interchange format -- equivalent to SCORM for courseware -- does not exist in the simulation training space.
Haptic hardware drives costs beyond most training budgets. Force-feedback devices required by platforms like FundamentalVR add thousands of dollars per station. A software-defined haptic approximation layer running on consumer VR controllers could dramatically lower the barrier to entry.
No cross-vertical scenario portability. A scenario authored for military training cannot be imported into an industrial safety platform without complete rework. Every platform is vertically siloed.
Static scenarios dominate. Most platforms deliver fixed branching trees rather than AI-driven adaptive difficulty. Real-time complication injection and coaching based on learner performance remains rare outside niche research prototypes.
Scenario authoring requires developers. Platforms like Osso VR and FundamentalVR have no documented SDK or no-code tooling for domain experts to create content. Training organisations are locked into vendor content libraries or expensive professional services engagements.

Key Features

Physics-Based Simulation Engine

Real-time rigid-body, soft-body, and fluid simulation for accurate tissue, equipment, and environmental force representation
Configurable fidelity levels balancing realism against hardware capability (full GPU physics on high-end hardware; reduced-fidelity approximation on standalone VR)
Haptic device abstraction layer supporting Touch X (OpenHaptics) and HaptX Gloves via a unified API
Cross-platform delivery: native VR (Meta Quest 3, Valve Index, PlayStation VR2), PC-based simulation, and mobile-compatible simplified scenarios

Scenario Authoring Toolkit

GUI-based editor for composing training scenarios from a library of asset primitives
Define branching decision points, complication injection (unexpected bleeding, equipment failure), and learning objectives
Scenarios stored in open JSON/YAML format enabling version control, collaborative editing, and cross-platform portability
Content marketplace with publisher tooling and revenue-sharing model for domain experts to monetise scenario packs

AI-Powered Adaptive Training

Real-time adaptive difficulty adjusting complication injection frequency and type based on rolling performance score
Conversational AI virtual patients and characters for communication and decision-making scenarios
Skill decay prediction with automated spaced-repetition reassessment scheduling
Under-pressure performance prediction from micro-behavioural signals (hesitation, gaze fixation, controller jitter)

Competency Assessment and Analytics

Map completed actions to configurable competency frameworks (ACGME milestones for surgery, OSHA standards for industrial safety)
Structured proficiency reports with per-session performance breakdown
Cohort-level analytics comparing performance across trainees, tracking skill decay over time, and flagging individuals not progressing
Automated competency gap detection recommending targeted remediation content

Integration and Standards Compliance

xAPI/cmi5 reporting to a Learning Record Store; SCORM 1.2 package export for legacy LMS compatibility
LTI 1.3 for direct LMS integration with Moodle, Cornerstone, or SAP SuccessFactors
OpenXR-compliant runtime as the hardware abstraction layer across VR device fleets
Multi-user collaborative simulation with server-authoritative physics state synchronisation

AI-Native Advantage

Unlike incumbent platforms that deliver static branching scenarios, this platform uses AI to dynamically adjust scenario complexity in real time -- modifying complication frequency, injecting new stressors, and providing step-by-step corrective coaching during and after each session. Conversational AI drives open-ended clinical interviews and communication scenarios rather than scripted dialogue trees. AI-powered physics parameter calibration tunes tissue-resistance values against expert feedback, maintaining clinical accuracy as new procedures are added. Behavioural signal analysis (gaze patterns, economy of movement, hesitation timing) enables predictive assessment of real-world readiness before a trainee is cleared for live practice.

Tech Stack & Deployment

The platform targets OpenXR 1.1 as its hardware abstraction layer, ensuring compatibility across consumer and enterprise VR headsets without vendor lock-in. Physics simulation leverages GPU-accelerated engines (NVIDIA PhysX/Flex) on compatible hardware, with a reduced-fidelity fallback for standalone headsets using pre-baked deformation sequences. Haptic feedback runs on a separate low-latency thread (sub-1 ms refresh) decoupled from the visual render loop. Multi-user scenarios use deterministic physics with a server-authoritative state model to prevent desynchronisation. Cloud rendering can be streamed to standalone headsets for high-fidelity scenes without local GPU constraints. Scenario data is stored in structured JSON/YAML graphs, keeping logic separate from assets so non-programmer domain experts can modify procedures without touching code.

Market Context

The global virtual training and simulation market was valued at $449.9 billion in 2024 and is projected to reach $844.2 billion by 2030 at an 11.1% CAGR, spanning defence, industrial safety, equipment operation, and healthcare. The VR medical training segment alone is projected at approximately $395.4 million in 2025, growing at nearly 16% annually through 2033. All major incumbents are proprietary commercial platforms with high per-seat licensing costs and closed content ecosystems. Studies report 42% improvement in procedural accuracy, 38% reduction in training time, 45% decrease in error rates, and 48% increase in trainee confidence compared to traditional methods. Primary buyers are hospital systems, medical residency programmes, industrial safety departments, defence training commands, and corporate L&D organisations.

Project Status

This project is in the research and specification phase.
Contributions, feedback, and domain expertise are welcome.

Contributing

We welcome contributions from developers, domain experts, and potential users. See CONTRIBUTING.md for guidelines.

Important: All contributions must be your own original work or clearly attributed open-source material with a compatible licence. Copyright infringement and licence violations will not be tolerated and will result in immediate removal of the offending contribution. If you are unsure whether a piece of code, text, or other material is safe to contribute, open an issue and ask before submitting.

Licence

Licence to be determined. See discussion for context.