NASA’s Perseverance rover completed its first AI-planned drives on Mars on December 8 and 10, 2025, using generative AI to navigate Jezero Crater’s rim without human input, according to NASA.
The Perseverance rover’s December 2025 drives marked a milestone in autonomous space exploration, with generative AI creating safe routes over Martian terrain. The mission, led by NASA’s Jet Propulsion Laboratory (JPL), used a vision-language model to analyze HiRISE imagery and terrain data, generating waypoints for the six-wheeled rover. This reduced reliance on human planners, who typically spaced waypoints no more than 330 feet (100 meters) apart to avoid hazards.
AI-Driven Navigation on Mars
Perseverance executed two AI-planned drives on sols 1,707 and 1,709, covering 689 feet (210 meters) and 807 feet (246 meters), respectively. The AI, developed in collaboration with Anthropic’s Claude models, processed high-resolution orbital imagery and digital elevation data to identify hazards like boulder fields and sand ripples. A “digital twin” of the rover verified over 500,000 telemetry variables before commands were sent to Mars, which is 140 million miles (225 million kilometers) from Earth on average.
The Role of Generative AI in Rover Planning
The AI’s analysis mirrored human planners’ workflows, using the same data sources to generate waypoints. This approach allowed Perseverance to adapt to terrain challenges without real-time human intervention. Autonomous technologies like this can help missions operate more efficiently, respond to challenging terrain, and increase science return as distance from Earth grows,
said NASA Administrator Jared Isaacman.
The collaboration with Anthropic highlighted the integration of commercial AI tools into space missions. The AI’s output was processed through JPL’s digital twin, a virtual replica of the rover, to ensure compatibility with its flight software. This step underscored the balance between innovation and safety, as even minor discrepancies in command execution could jeopardize the mission.
Precision Engineering of the Roman Telescope’s Mirror
While Perseverance’s AI marked a leap in autonomy, NASA’s Nancy Grace Roman Space Telescope achieved a critical milestone in May 2026. Engineers at Goddard Space Flight Center completed the final inspection of its 7.9-foot (2.4-meter) primary mirror, which will collect light for cosmic observations. The mirror’s surface is smoother than required, with bumps no taller than 1.2 nanometers—equivalent to a quarter-inch on a globe-sized scale.
The mirror’s silver layer is less than 400 nanometers thick, optimizes near-infrared light reflection, a key feature for the telescope’s dark matter and exoplanet studies. The mirror passed with flying colors, keeping the mission on track for an early September launch,
said Bente Eegholm, optics lead. The telescope’s ultralow-expansion glass ensures stability during temperature shifts, preserving image quality in space.
Implications for Future Space Missions
The Perseverance AI demonstration and Roman telescope’s mirror inspection reflect broader trends in space exploration: increased reliance on autonomous systems and precision engineering. For Mars missions, autonomous navigation could enable longer, more complex traversals as communication delays persist. For telescopes, the Roman mirror’s specifications set a new standard for optical precision, critical for deep-space imaging.
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.
Challenges in Autonomous Space Exploration
Despite these advancements, challenges remain. The vast distance to Mars limits real-time control, necessitating autonomous systems. For the Roman telescope, the mirror’s inspection process—using a high-resolution camera to verify alignment—highlighted the complexity of ensuring precision in space. In order to gather very sensitive measurements of objects strewn throughout space, all of Roman’s components have to be ultraprecise,
said Eegholm. The primary mirror certainly delivers on that precision.
As NASA prepares for the Roman telescope’s September launch and future Mars missions, the integration of AI and precision engineering will shape how humanity explores space. These developments signal a shift toward systems that can operate independently, adapt to unknown environments, and push the boundaries of scientific discovery.