What is NASA’s ExoMiner++? How AI is scanning space data and helping to find new planets

NASA is using artificial intelligence to speed up the search for planets beyond our solar system. The latest tool, called ExoMiner++, helps scientists study huge amounts of space data and identify possible new exoplanets more quickly and accurately.

Developed by researchers at NASA’s Ames Research Center, ExoMiner++ builds on earlier AI work and uses data from two major space missions Kepler and TESS.

What is ExoMiner++

ExoMiner++ is an AI-based software tool designed to analyse transit photometry data. This data comes from tiny dips in a star’s brightness when a planet passes in front of it, known as a transit.

According to NASA, the goal of the ExoMiner project is to automatically sort through hundreds of thousands of such signals and flag the ones that are most likely caused by planets. Many signals turn out to be noise or other space events, such as eclipsing binary stars.

The original ExoMiner, created in 2021, helped confirm 370 new exoplanets using Kepler data. The new version, ExoMiner++, has been trained using both Kepler and TESS data, making it more flexible and powerful.

How ExoMiner++ uses AI to find planets

ExoMiner++ uses machine learning to classify transit signals and decide whether they are likely to be planets. In an early test run using TESS data, the tool flagged around 7,000 targets as potential exoplanet candidates.

NASA scientists explain that an exoplanet candidate is not a confirmed planet. These signals still need follow-up observations from other telescopes. However, AI helps narrow down the list, saving scientists years of manual work.

Miguel Martinho, a co-investigator on the project at NASA Ames, said AI is ideal for this task because of the sheer volume of data involved. According to him, when there are hundreds of thousands of signals, deep learning tools become essential.

Data from Kepler and TESS

Kepler and TESS observe space in different ways. Kepler studied a small area of the sky in great detail, while TESS scans almost the entire sky, focusing on nearby stars.

Despite these differences, NASA says the data from both missions is compatible. This allows ExoMiner++ to learn from both datasets and deliver more reliable results. Project lead Hamed Valizadegan noted that combining data from the two missions allows strong scientific returns with limited resources.

What comes next

ExoMiner++ is open-source and available on GitHub. This means researchers around the world can download it, test it, and use it on public space data.

Kevin Murphy, NASA’s chief science data officer, said open-source tools help scientists reproduce results and explore data more deeply. He added that open data and shared code are key parts of modern scientific research.

NASA researchers are already working on the next version of ExoMiner++. In the future, the tool may not just analyse transit signals but also detect them directly from raw telescope data.

Jon Jenkins, an exoplanet scientist at NASA Ames, said NASA’s open science approach is improving both research and software. According to him, open-source science is one of the main reasons the field of exoplanet discovery is moving so fast.

As more space data becomes available, tools like ExoMiner++ are expected to play a central role in finding new worlds beyond Earth.