CDM Cross-Validation Study
We tested OrbVeil against 100 official Conjunction Data Messages from the 18th Space Defense Squadron. Here's exactly how we performed — and where we fall short.
We obtained 100 Conjunction Data Messages (CDMs) from Space-Track.org, generated by the 18th Space Defense Squadron using high-fidelity orbital data. We then ran OrbVeil's TLE-based conjunction screening on the same object pairs and compared results.
100 real conjunction events with precise miss distances and TCAs computed from SP ephemerides by the US Space Force.
For each CDM pair, we ran OrbVeil using publicly available TLEs. Same objects, same time windows, independent computation.
Compared OrbVeil's miss distance and TCA against the CDM ground truth. Measured absolute errors and success/failure rates.
TLE-based screening is fundamentally different from the high-fidelity methods used by military space surveillance. We believe in transparency about what OrbVeil can and cannot do.
OrbVeil is not a replacement for operational collision avoidance decisions. Maneuver planning requires SP ephemerides, covariance data, and human-in-the-loop processes that TLE screening cannot provide.
Expect miss distance errors in the 1–10 km range. TLEs are general perturbation fits — they lack the precision of special perturbation ephemerides used in official CDMs.
Accuracy degrades with TLE age. Our test set averaged 4.0-day-old TLEs. Maneuvering satellites or stale TLEs will produce larger errors.
We compute miss distance, not collision probability. Without covariance matrices, we cannot estimate the true likelihood of collision — only geometric proximity.
OrbVeil is designed for awareness and first-pass screening — not final operational decisions.
OrbVeil is open-source. Inspect the algorithms, run your own validation, or request a threat report for your constellation.