Did You Know NASA Has Been Using AI For Decades?

Did you know that NASA has been using AI for decades? Long before ChatGPT made headlines, AI tools were already integrated into NASA missions and research. AI is being used in a growing number of key operational areas in space, including autonomous rover driving, planetary data analysis, and mission research and planning.

Let’s take a look at a couple specific use cases.

Navigating Mars: AI on the Perseverance Rover

The Mars rover Perseverance, which landed in February 2021, relies on an autonomous navigation system called AutoNav to drive itself across the Martian surface. This system allows the rover to assess its environment, plan routes, and avoid hazards without needing constant input from mission controllers.

Autonomy is a critical feature for the safety of the rover when communication lags with rover controllers occur. They can last several minutes.

In mid-2024, Perseverance crossed a particularly rocky region known as Snowdrift Peak. Why is this relevant? The planned path through this area was about 520 meters (1,706 feet), but the rover ultimately traveled 759 meters (2,490 feet) as it rerouted around unexpected obstacles that hadn’t appeared in satellite imagery. The autonomous portion of the drive, excluding science stops, took just six Martian days (sols) — significantly faster than previous rovers like Curiosity would have managed on similar terrain. According to NASA, mission planners estimate that this route would have taken Curiosity about 12 additional sols to complete, highlighting the improvements in autonomous mobility between missions.

Perseverance includes a second onboard computer dedicated to image processing, enabling it to analyze terrain in real time and drive without stopping to process data. This hardware-software combination makes faster and safer autonomous navigation possible — especially in situations where there is no immediate feedback from Earth.

These capabilities build on decades of iterative progress, starting with Sojourner in 1997 and continuing through Spirit, Opportunity, and Curiosity. Perseverance represents the most autonomous surface operations NASA has deployed to date.

Confirming Exoplanets: The Role of ExoMiner

Another great example is ExoMiner, a deep learning model developed to help NASA sift through archival space telescope data and confirm potential exoplanets. Rather than collecting new data, ExoMiner revisits observations made by the Kepler and K2 missions, which recorded brightness levels of stars between 2009 and 2018.

The model is trained to recognize patterns consistent with transiting exoplanets — subtle dips in a star’s brightness that suggest a planet may have passed in front of it. ExoMiner applies this logic across large datasets to flag strong candidates and distinguish them from false positives like binary star activity or noise.

In one round of analysis, ExoMiner helped confirm 301 new exoplanets that had previously been listed as candidates but never officially verified. These findings were based on existing data, demonstrating how machine learning can unlock new scientific value from historical archives. Scientists then cross-checked the model’s results before confirming each planet.

Looking Ahead: Commercial Applications for AI in Space

The AI systems currently used by NASA offer a starting point for broader adoption in the commercial space sector. As private companies take on more ambitious missions, the ability to automate decision-making, process large datasets in real time, and operate in cases where human input is delayed will likely be explored and incorporated.