DE2Sim

🔴 Live Demo — dbbun.github.io/DE2Sim

Runs entirely in your browser — dashboard, 3D geometry viewer, simulation viewer, ASOT traceability, and behavior review. No install, no server, no internet required after the page loads.

DE2Sim converts standards-based digital-engineering artifacts (CAD-export geometry, a SysMLv2 subset, parametric data) into an Authoritative Source of Truth (ASOT), AI-assisted behavior proposals reviewed and approved by a human, and executable low- and high-fidelity simulation output — packaged as a self-contained, offline-runnable browser demo.

Originally developed for the Army Training Verse Prize Challenge II ("Intelligent Simulation Pipeline").

Getting the Code

git clone https://github.com/DBbun/DE2Sim.git
cd DE2Sim

To pick up later changes in an existing clone:

git pull origin master

Setup

• Python 3.10 or later. The core de2sim package is standard-library only — no pip install required to run the pipeline itself.
• The legacy paper_to_simulator_builder_v3_4.py script referenced in earlier project history has been removed from this repository; it is not part of the current de2sim pipeline and is not needed to run anything below.
• Optional: to use a local generative-AI provider for behavior enrichment (--ai-provider ollama), install and run Ollama separately and pull a model (e.g. ollama pull gemma3:4b). Everything else, including the default --ai-provider offline mode, needs no external services.

Quickstart

The repo ships a working example engineering package, so you can run the full pipeline immediately without preparing your own input:

python -m de2sim.cli.challenge_pipeline \
  --engineering-package demo_outputs/canonical/engineering_package/DBbun-DE2Sim-DemoUAS-Geometry.zip \
  --output out --build-asot
python -m de2sim.cli.challenge_pipeline --output out --propose-behaviors
python -m de2sim.cli.challenge_pipeline --output out --apply-behavior-decisions out/behavior_decisions.json
python -m de2sim.cli.challenge_pipeline --output out --build-simulation
python -m de2sim.cli.challenge_pipeline --output out --build-demo-package

The last step writes a self-contained package under out/ — open its demo_dashboard.html in any browser to see the result (no server, no internet connection needed). To skip running the pipeline yourself entirely, the same kind of output is already built and viewable at the live demo or under demo_outputs/canonical/submission_package/.

Running Tests

python -m unittest discover -s tests

Layout

• de2sim/ — the pipeline package (ingest, geometry, ASOT, provenance, behaviors, simulation, visualization, CLI, demo packaging).
• docs/ — architecture notes and phase-by-phase build documentation.
• tests/ — unit tests for each pipeline stage.
• demo_outputs/canonical/ — canonical example run: the source engineering package (also used by the Quickstart above), local-AI/ASOT output, simulation output, and the final packaged submission demo (open submission_package/DE2Sim_Submission_Demo.zip, extract, and open demo_dashboard.html — no Python, internet, or GPU required to view it). The interactive simulation_viewer.html lives only under submission_package/.../viewers/ to avoid duplicating an ~4 MB file; simulation_output/ keeps the underlying JSON/CSV data without a second copy of the viewer.

How to Add Engineering Data

DE2Sim is command-line only — there is no upload UI. You provide an engineering package: a ZIP file laid out so de2sim/ingest/ can classify its contents by folder name:

your_package.zip
├── geometry/        → .stl, .glb, .gltf, .obj
├── sysml/           → .sysml files (SysMLv2)
├── parameters/      → .csv, .json, .yaml, .xlsx
├── physical_models/ → .md, .txt, .json, .yaml
└── requirements/    → .csv, .json, .yaml, .md, .txt

Then run the CLI, chaining stages as needed:

python -m de2sim.cli.challenge_pipeline --engineering-package package.zip --output out --build-asot
python -m de2sim.cli.challenge_pipeline --output out --propose-behaviors
python -m de2sim.cli.challenge_pipeline --output out --apply-behavior-decisions out/behavior_decisions.json
python -m de2sim.cli.challenge_pipeline --output out --build-simulation
python -m de2sim.cli.challenge_pipeline --output out --build-demo-package

Caveat: only structured formats are actually parsed for content — .json/.yaml fully, and .md/.txt in physical_models/ only for lines shaped like name = equation or equation: .... .pdf/.docx files are accepted and referenced for provenance, but their content is not extracted — physical_model_reader.py explicitly returns "reader does not support" for those extensions. Geometry files are likewise referenced at ingestion and only actually parsed/validated in the later --extract-geometry stage. In short: bring structured data (CSV/JSON/YAML/ STL/SysML), not a raw spec document you expect the pipeline to read for you.

License

Licensed under the DBbun Source-Available License v1.0 (see LICENSE): free for genuine academic research use. Any commercial or government use — including evaluation, internal deployment, or use under a government contract, grant, or prize challenge — requires a separate license from DBbun LLC. Contact contact@dbbun.com.