AI-Coordinated Space Optical Interferometry for Direct Imaging of Exoplanet Surfaces
Can we photograph continents on a planet orbiting another star? We believe the answer is yes — with a formation-flying swarm of small satellites acting as a single 100+ km optical telescope, coordinated and reconstructed by AI.
- Distributed aperture: 10–20 small satellites in precision formation flight → optical interferometer with 100–1000 km baselines
- Dynamic hierarchical nulling: suppressing host-star light to reveal the planet
- AI-driven reconstruction: recovering surface maps from extremely sparse uv-plane data (~2% coverage)
At λ = 550 nm and a 100 km baseline, the diffraction limit at 1.3 pc (Alpha/Proxima Centauri) corresponds to ~250 km surface resolution — enough to map continents, oceans, and ice caps.
Our uv-plane simulation (10 satellites, 100 km circular array, 24-epoch super-synthesis) achieves only ~1.8% uv coverage — demonstrating why classical imaging fails and AI-based sparse reconstruction is essential.
Left to right: ground-truth surface model, uv coverage, dirty image, Wiener reconstruction.
| Scenario A: Alpha Cen A + Earth-like | Scenario B: Proxima b (real planet) | |
|---|---|---|
| Star | G2V, V=0.01 (bright) | M5.5V, V=11.13 (~26,000× fainter) |
| Contrast | ~10⁻¹⁰ | ~8×10⁻⁸ (~760× better) |
| Regime | Photon-rich, null-limited | Photon-limited, null-relaxed (~10⁻⁶) |
| Separation | ~930 mas | ~37 mas (harder IWA) |
Full analysis: docs/technical_overview.md
simulations/ # Python simulation code (baseline analysis, uv-plane synthesis)
figures/ # Generated figures
docs/ # Technical overview
git clone https://github.com/mapping-alien-worlds/mapping-alien-worlds.git
cd mapping-alien-worlds
pip install numpy scipy matplotlib
python simulations/uv_plane_sim.py- NASA NIAC 2026 — Paul Stankus (Brookhaven), "Mapping Alien Continents: Achieving Optical VLBI for Exoplanet Imaging"
- LIFE (Large Interferometer For Exoplanets) — ETH Zürich, mid-IR nulling formation-flying concept
- SILVIA (JAXA) — precision formation-flying demonstration
- Baseline–resolution feasibility analysis
- Sparse uv-plane synthesis simulation
- Two-scenario photon budget (Alpha Cen A vs Proxima b)
- AI reconstruction benchmark (diffusion prior vs. CLEAN/Wiener)
- End-to-end photon budget & nulling model
- Formation-flight OPD stabilization requirements study
- Community white paper
We welcome astronomers, ML researchers, and space systems engineers. Open an issue or start a discussion — see CONTRIBUTING.md
MIT License — free to use, modify, and distribute with attribution.
Gökhan Can — Concept originator & project lead
Open a GitHub Issue or reach out via Discussions.
© 2026 Gökhan Can — MIT License