AI Finds 100+ Hidden Planets in NASA Data
Astronomers at the University of Warwick developed a new AI tool. Its name is RAVEN. The system analyzed data from NASA’s TESS mission. It scanned over 2.2 million stars. As a result, RAVEN confirmed 118 new planets. Thirty‑one of them are completely new discoveries.
Rare and Extreme Worlds
Some of these planets orbit their stars in less than 24 hours. Astronomers call them ultra‑short‑period planets. Others lie in the so‑called “Neptunian desert.” This is a region where planets were thought to be scarce. The AI also found tightly packed multi‑planet systems. For example, several pairs of planets orbit the same star.
How RAVEN Works
Many false signals can mimic planets. Eclipsing binary stars are a common example. RAVEN uses machine learning trained on realistic simulations. It identifies genuine planets with high accuracy. In addition, the pipeline handles the whole process in one go. It detects signals, vets them, and validates the results.
Measuring Planet Populations
The team also measured how common close‑in planets are. About 9‑10% of Sun‑like stars host such a world. Neptunian desert planets appear around only 0.08% of Sun‑like stars. This is the first precise measurement of its kind. “RAVEN allows us to analyze enormous datasets consistently,” says Dr. David Armstrong.
A New Era for Astronomy
The team released interactive catalogs for other scientists. Future missions like ESA’s PLATO will benefit.AI is transforming planet hunting. The galaxy just became a lot more crowded.
