AI System RAVEN Discovers 118 New Exoplanets Including Rare Extreme Worlds
Astronomers at the University of Warwick have confirmed more than 100 exoplanets—including 31 entirely new worlds—using an advanced artificial intelligence system called RAVEN. The breakthrough demonstrates how machine learning is transforming astronomical discovery, turning massive datasets into reliable planetary detections with unprecedented precision.
The Discovery
The team applied the RAVEN pipeline to data from NASA's Transiting Exoplanet Survey Satellite (TESS), analyzing observations from over 2.2 million stars gathered during the mission's first four years. Their research, published in the Monthly Notices of the Royal Astronomical Society (MNRAS), focused on planets orbiting very close to their host stars—completing full orbits in less than 16 days.
"Using our newly developed RAVEN pipeline, we were able to validate 118 new planets, and over 2,000 high-quality planet candidates, nearly 1,000 of them entirely new," said Dr. Marina Lafarga Magro, Postdoctoral Researcher at the University of Warwick and first author of the study. "This represents one of the best characterized samples of close-in planets and will help us identify the most promising systems for future study."
Rare and Extreme Worlds Identified
Among the confirmed planets are several particularly intriguing categories:
- Ultra-Short-Period Planets: Worlds that circle their stars in under 24 hours, experiencing extreme temperatures and tidal forces - 'Neptunian Desert' Planets: Rare planets occupying a region where few should exist based on current formation theories - Tightly Packed Multi-Planet Systems: Previously unknown pairs of planets orbiting the same star in close proximity
The discovery of Neptunian desert planets is particularly significant. These worlds occupy an orbital zone where intense stellar radiation should strip away planetary atmospheres, making their existence puzzling to astronomers.
How RAVEN Works
Modern planet-hunting missions like TESS flag thousands of possible planets, but distinguishing genuine planetary signals from false positives remains a major challenge. Eclipsing binary stars, instrument noise, and other astrophysical phenomena can mimic the dimming patterns caused by transiting planets.
"The challenge lies in identifying if the dimming is indeed caused by a planet in orbit around the star or by something else, like eclipsing binary stars, which is what RAVEN tries to answer," explained Dr. Andreas Hadjigeorghiou of Warwick, who led the pipeline's development. "Its strength stems from our carefully created dataset of hundreds of thousands of realistically simulated planets and other astrophysical events that can masquerade as planets. We trained machine learning models to identify patterns in the data that can tell us the type of event we have detected—something that AI models excel at."
Unlike contemporary tools that focus on specific parts of the detection workflow, RAVEN handles the entire process end-to-end: from initial signal detection, through machine learning vetting, to statistical validation. This integrated approach gives the pipeline a significant advantage in accuracy and efficiency.
Measuring Planetary Prevalence
Beyond individual discoveries, the validated dataset enabled the researchers to examine broader patterns in planetary populations. In a companion MNRAS study, they measured how often close-in planets occur around Sun-like stars, mapping results by orbital period and planet size with unprecedented detail.
Key findings include:
- 9-10% of Sun-like stars host a close-in planet, aligning with earlier Kepler mission findings but with uncertainties reduced by up to a factor of ten - 0.08% of Sun-like stars host 'Neptunian desert' planets—the first direct measurement of just how rare these worlds are
"For the first time, we can put a precise number on just how empty this 'desert' is," said Dr. Kaiming Cui, Postdoctoral Researcher at Warwick and first author of the population study. "These measurements show that TESS can now match, and in some cases surpass, Kepler for studying planetary populations."
Why This Matters
The RAVEN pipeline represents a paradigm shift in astronomical research. By combining massive datasets with machine learning, researchers can now:
- Accelerate Discovery: Process enormous volumes of telescope data far faster than manual analysis allows - Reduce False Positives: AI-trained models filter out noise and astrophysical impostors with high accuracy - Correct for Detection Biases: The system evaluates which types of planets are easier or harder to detect, enabling more accurate population statistics - Enable Follow-Up Studies: Reliable catalogs help other scientists identify promising targets for ground-based telescopes and future missions like ESA's PLATO
Dr. David Armstrong, Associate Professor at Warwick and senior co-author, emphasized the reliability of the results: "RAVEN allows us to analyse enormous datasets consistently and objectively. Because the pipeline is well-tested and carefully validated, this is not just a list of potential planets—it is also reliable enough to use as a sample to map the prevalence of distinct types of planets around Sun-like stars."
Open Data for the Scientific Community
The team has released interactive catalogs and tools, allowing other scientists to explore the results and identify promising targets for follow-up observations. This open-science approach accelerates collaborative research and maximizes the scientific return from TESS data.
The Future of AI-Driven Astronomy
These studies highlight how artificial intelligence is becoming indispensable in modern astronomy. As telescope missions generate ever-larger datasets—from TESS to the upcoming Vera C. Rubin Observatory and beyond—machine learning systems like RAVEN will be essential for turning raw observations into discoveries.
The success of RAVEN also points toward broader applications: similar AI pipelines could be adapted for other astronomical challenges, from gravitational wave detection to galaxy classification. The era of AI-assisted planetary science has arrived, and it's already rewriting our understanding of how common different types of worlds are across the galaxy.
What Is RAVEN?
RAVEN (Robotic Automated Validation of Exoplanets with Neural networks) is an automated system designed to address one of astronomy's biggest challenges: converting enormous volumes of space telescope data into reliable discoveries. It scans data from millions of stars to find the tiny drops in brightness caused by planets passing in front of them, then uses AI trained on realistic simulations to filter out false signals before statistically confirming the strongest candidates.
Importantly, RAVEN also evaluates detection biases, helping researchers correct for hidden systematic errors. This means it not only speeds up the discovery of new worlds but also produces cleaner, more reliable datasets for answering fundamental questions about planetary formation and prevalence.
