NASA announced a major milestone in space exploration as the Perseverance rover completed its first drive on Mars with a route designed entirely by Generative AI tools . The test took place on December 8 and 10 , 2025, and marks the first time an AI system has autonomously planned navigation in such a demanding extraplanetary environment.
This demonstration shows that artificial intelligence can analyze the rough terrain of Mars and suggest safe routes without the need for manual intervention by engineers on Earth, automating a process that until now required enormous human effort.
NASA opens a new chapter in autonomous exploration
NASA Administrator Jared Isaacmancalled the effort a testament to the technological progress the agency is making.
He said such autonomous technologies could allow future missions to operate more efficiently, respond better to hazardous locations, and increase scientific performance as spacecraft travel further and further from Earth.
See also: Mars was once blue – New evidence for a lost ocean
Perseverance is already one of the most advanced rovers ever sent to Mars, but the integration of Generative AI turns it into a precursor to a new era where robots will make more decisions on their own.
Why autonomy is essential on Mars
Mars is about 225 million kilometers away from Earth on average. This means that communications experience delays that make real-time control impossible.
For decades, mission teams have been plotting rover routes manually every day. Engineers analyze terrain images, slope data, and vehicle condition, then map out routes with small intermediate waypoints, usually no larger than 100 meters, to avoid rocks, sand traps, or steep slopes.
Commands are sent via Deep Space Network , and the rover executes them several hours later. It's a slow but necessary process, but one that limits the speed of exploration.
JPL and Anthropic Collaboration: Claude AI at the helm
The artificial intelligence test drive was developed at NASA's Jet Propulsion Laboratory (JPL) in California, in collaboration with the company Anthropic , using models Claude AI .
See also: NASA Artemis 2: Astronauts quarantined shortly before launch
The artificial intelligence analyzed exactly the same data that human designers use: images from the Mars Reconnaissance Orbiter, as well as computer models that provide information about the slope and hazard of the terrain.
From this information, the AI system identified critical surface features, such as cliffs, steep slopes, and rocks, and plotted a course with specific navigation points that the rover had to follow sequentially.
Perseverance: 456 meters to Jezero Crater in complete safety
During the two test runs, Perseverance traveled approximately 456 meters along the rim of Jezero Crater.
Before commands were sent to Mars, the mission team extensively tested the designs through a detailed “digital twin” of the rover, a simulated model that confirms that the movements can be executed safely.
This particular step is critical, as any mistake on Mars could cost an entire mission billions.
The future: Rovers exploring kilometers without human guidance
Vandi Verma, a space robotics specialist at JPL, emphasized that the foundations of Generative AI offer enormous potential in autonomous off-planet navigation .
According to her, we are approaching an era where rovers will be able to cover kilometers of travel with minimal human intervention, reducing the workload of operators and boosting scientific output.
See also: AI found 1,400 'anomalous objects' from Hubble archives
At the same time, such tools will be able to automatically highlight geological features worthy of study, analyzing huge volumes of images much faster than any human team.
Artificial intelligence becomes NASA's new "co-pilot"
The success of Perseverance's AI journey is not just a technical demonstration. It's a first step toward missions where artificial intelligence will serve as a key partner in the exploration of other worlds, paving the way for faster, safer, and more efficient space exploration.
