NASA’s Perseverance Rover Completes First AI-Planned Drive on Mars
Perseverance completes first AI-planned drive on Mars
NASA’s Perseverance rover has achieved a historic milestone on Mars, successfully completing the first drives on another planet planned entirely by artificial intelligence.
The breakthrough demonstration, led by NASA’s Jet Propulsion Laboratory (JPL) in Southern California, took place on December 8 and 10. During the test, a vision-enabled generative AI system selected the rover’s driving waypoints a task traditionally performed by human mission planners on Earth.
The achievement signals a future in which robotic explorers can operate more independently, reducing reliance on Earth-based teams and enabling more efficient deep-space missions.
AI Takes the Wheel
For nearly three decades, Mars rover navigation has depended on expert human planners who analyze terrain images and carefully map safe routes. Because Mars is about 140 million miles (225 million kilometers) from Earth on average, communication delays make real-time control impossible.
Instead, engineers design paths made up of short segments usually no more than 330 feet (100 meters) — to minimize the risk of encountering hazards such as rocks, steep slopes, or soft sand. These instructions are transmitted via NASA’s Deep Space Network, and the rover executes them autonomously.
In the recent test, that responsibility shifted to artificial intelligence.
Engineers used a type of generative AI known as a vision-language model to analyze high-resolution orbital images captured by the HiRISE camera aboard NASA’s Mars Reconnaissance Orbiter, along with detailed terrain slope data. The AI examined the same datasets human planners use, identifying bedrock, boulder fields, sand ripples, and other surface features before generating a continuous driving path complete with necessary waypoints.
The project was coordinated from JPL’s Rover Operations Center in collaboration with AI company Anthropic, using its Claude models.
Extensive Safety Testing
Before transmitting the AI-generated route to Mars, engineers validated the plan using JPL’s digital twin a virtual replica of the rover. The system checked more than 500,000 telemetry variables to ensure the commands would operate safely within Perseverance’s flight software.
Once cleared, the rover executed the AI-planned drives.
On December 8, Perseverance traveled 689 feet (210 meters). Two days later, it completed another 807 feet (246 meters), successfully following the AI-generated routes across challenging Martian terrain.
A New Era of Autonomous Exploration
“This demonstration shows how far our capabilities have advanced and broadens how we will explore other worlds,” said NASA Administrator Jared Isaacman. “Autonomous technologies like this can help missions operate more efficiently, respond to challenging terrain, and increase science return as distance from Earth grows.”
Vandi Verma, a space roboticist at JPL and member of the Perseverance engineering team, said generative AI is showing strong potential in streamlining the key components of autonomous navigation: perception, localization, and planning.
“We are moving towards a day where generative AI and other smart tools will help our surface rovers handle kilometer-scale drives while minimizing operator workload,” Verma said.
NASA officials believe such technology could become essential for future missions including sustained human presence on the Moon and eventual crewed missions to Mars.
“Imagine intelligent systems not only on the ground at Earth, but also in edge applications in our rovers, helicopters, drones, and other surface elements trained with the collective wisdom of our NASA engineers, scientists, and astronauts,” said Matt Wallace, manager of JPL’s Exploration Systems Office. “That is the game-changing technology we need to establish the infrastructure and systems required for a permanent human presence on the Moon and take the U.S. to Mars and beyond.”
With Perseverance’s successful AI-planned drive, that future may be closer than ever.
