Tick-based agent-based model of maritime traffic safety on the Baltic and North Sea.

Vessels follow depth- and lane-constrained routes, resolve encounters with a CPA/COLREGs give-way rule gated by weather × crew fatigue, and — on collision — feed DTU SIMCOL damage plus an IAMSAR/MASSIM + Ashrafi SAR chain. Batches compare methods and scenarios with non-parametric hypothesis tests.

# Terminal 1 — API (from repo root; reads data/, writes outputs/)
cargo run -p api --release

# Terminal 2 — UI
cd frontend && npm install && npm run dev

Open http://localhost:5173. Launch a small batch under Runs, pick a finished seed, scrub the map.

Prerequisites: Rust (stable), Node.js 20+. Optional: Python 3.11+ for tools/, LaTeX for the paper.

maritime-abm/
├── crates/simulation/   # ABM engine
├── crates/api/          # HTTP batch API + SSE
├── crates/aisprocess/   # depth/lane route generator
├── frontend/            # playback workbench
├── data/                # committed calibration inputs
├── tools/               # fetchers, sweep, plotting
├── docs/                # implementation docs
├── report/              # academic write-up
└── outputs/             # gitignored batch artifacts

data/ is the calibration baseline — do not overwrite casually. See docs/data-and-routes.md.

An earlier Python / Streamlit prototype lives in the archived repo Pawlo77/maritime-safety-simulation. This Rust + React codebase replaces it.

Course / research project at Warsaw University of Technology. See authors in report/report.pdf.