AI-LEO Policy Simulator
Non-operational

This tool is for research, policy analysis, and education only. It is not operational SSA, a collision-warning system, a maneuver-command tool, or flight-safety decision support.

AI-LEO Policy Simulator

A research-grade SGP4, policy-simulation, and AI-triage platform for multi-operator LEO governance analysis.

View static demo Quick tutorial Downloads Screenshots Run locally Cite and reuse

Why multi-operator LEO governance matters

Low Earth orbit is becoming increasingly crowded as large constellations and multiple operators place more satellites into overlapping orbital environments. The governance problem is not only technical screening; it is also coordination burden, responsibility allocation, and fairness. When close-approach-like situations occur, who should act, who carries the workload, and how should responsibility be distributed across operators under uncertainty?

Understand it in 10 seconds

Generate synthetic LEO satellites

Propagate orbits with SGP4

Screen close-approach candidates

Compare R0/R1/R2 governance regimes

Rank unresolved coordination actions with bounded AI

The simulator turns orbital-growth scenarios into reproducible governance evidence, not operational alerts.

Start here

1) Understand the 5-step workflow

Read the 10-second overview and the research workflow diagram.

2) Compare R0/R1/R2

See the “Governance regimes compared” section below.

3) Open the demo preview

Curated 1000-satellite static preview (orbits, candidates, R0/R1/R2, AI triage).

Open demo preview

4) Download quickstart

Grab copy-paste local-run commands and public-facing notes.

Open downloads

5) Cite DOI when released

Prefer DOI-backed software and dataset records for citation.

See citation plan

Actual simulator preview

A public static demo preview is available using curated 1000-satellite outputs. It shows orbit samples, conjunction-screening candidates, R0/R1/R2 policy comparison, operator burden, and AI triage without exposing the live backend.

Open static demo preview Read tutorial View screenshots

Screenshots

Overview of the Phase-C workflow: landing page, orbit visualization, conjunction-screening candidates, R0/R1/R2 comparison, bounded AI triage, and evidence/reproducibility.

Overview collage of the AI-LEO Policy Simulator workflow.
View screenshots

Governance regimes compared

These regimes are simplified policy abstractions for research comparison, not real legal rules or operational maneuver procedures.

R0 — Uncoordinated baseline
Each operator acts independently. The simulator uses this as a baseline to estimate what unresolved coordination risk looks like when there is no shared coordination rule.

R1 — Soft / capability-based coordination
Responsibility is assigned mainly to the operator or satellite with stronger action capability. This represents a cooperative coordination regime without explicit fairness enforcement.

R2 — Fairness-aware allocation
Responsibility considers both action capability and accumulated operator burden. This regime tests whether governance can reduce unfair concentration of responsibility while still managing residual risk.

Regime Meaning Main question
R0 Uncoordinated baseline What happens without shared coordination?
R1 Soft / capability-based coordination Does assigning the most capable actor reduce unresolved actions?
R2 Fairness-aware allocation Can burden be distributed more fairly without increasing residual risk too much?

Key capabilities

Synthetic LEO population generation

Configurable synthetic populations for scenario studies.

SGP4 propagation

Standardized orbit propagation for research time series.

Conjunction-candidate screening

Geometric close-approach screening with transparent proxies.

Governance regime comparison

Compare R0, R1, and R2 proxy regimes on the same candidates.

AI dataset export

Structured exports for reproducible learning experiments.

Bounded AI triage

Prioritization support for analyst review workflows only.

FastAPI backend

Local read-first API for dashboard consumption.

Next.js dashboard

Local visualization of orbits, policy metrics, and evidence.

Reproducible evidence pack

Figures, JSON manifests, and documented regeneration commands.

How the simulator will be available

  1. Public static website
    Explains the workflow, shows screenshots, and provides quickstart files.
  2. Static demo dashboard
    A future public demo can show pre-generated simulator outputs without exposing the backend.
  3. Local research simulator
    Researchers can run the full backend and dashboard locally from the public repository after cleanup.

The live public backend is intentionally deferred until security review, rate limits, and job-queue controls are added.

How to use the simulator locally

  1. Start backend
  2. Start dashboard
  3. Click “Run 100-sat demo”
  4. Explore Orbits, Conjunctions, Policy, AI Triage, and Evidence pages

From the repository root (terminal 1): 

python -m uvicorn phaseC.backend.main:app --host 127.0.0.1 --port 8000

Terminal 2: 

cd phaseC/frontend
npm install
npm run dev

Open: http://localhost:3000

Research workflow

Dashboard pages

Evidence pack

The Phase-C prototype produces a reproducible evidence trail: dashboard exports, figure manifests, and command-line regeneration steps documented in the repository’s Phase C materials. When the repository is cleaned for public release, curated archives may be linked with DOIs (see citation section). This static page does not host downloadable bundles; it describes the workflow only.

Cite and reuse

For formal citation, cite the SSRN paper now. After public release, cite the Zenodo software DOI and the Zenodo/Figshare evidence-pack DOI. Website pages improve visibility; DOI-backed archives remain the durable citation targets.

Current citation status

SSRN page: https://ssrn.com/abstract=6679959