NASA's Perseverance rover has achieved a historic milestone: its first AI-planned drive on another planet. On December 8th and 10th, 2025, commands generated by Anthropic's Claude AI guided the rover approximately 400 meters across the Martian surface.
The inherent communication delay between Earth and Mars, which can take up to twenty minutes each way, necessitates a degree of rover autonomy. Human operators cannot micromanage drives in real-time. Instead, they meticulously plan routes, send instructions, and later review the rover's actions. This instance marks the first time an AI has taken on this critical planning role.
Engineers at NASA's Jet Propulsion Laboratory (JPL) utilized Claude to plot a path through a rocky Martian terrain. Claude, the same AI model used for tasks like drafting emails and analyzing finances, demonstrated its versatility in complex Mars exploration technology.
AI navigates Martian obstacles
Perseverance, a car-sized robot equipped with advanced scientific instruments, has been active on Mars since its February 2021 landing. Its mission focuses on understanding the planet's geology, past climate, and collecting samples, potentially paving the way for future human missions. The rover's landing site in Jezero crater was selected for its past evidence of water, a key indicator for potential ancient microbial life.
Driving on Mars is inherently risky. A misstep can lead to the rover sliding, tipping, becoming immobilized, or getting stuck. Historically, human experts have painstakingly created a "breadcrumb trail" of waypoints using orbital and onboard camera imagery. This process is time-consuming and requires extensive expertise.
The daily brief
Get the free daily brief with the day's AI news, new AI startups and funding rounds every morning.
Join 64,000 other subscribers.
JPL engineers aimed to streamline this process by testing Claude's ability to plan routes with human-level accuracy. They provided Claude Code with extensive data and operational experience from years of rover driving.
Claude's Martian route planning
Claude generated commands in Rover Markup Language, an XML-based programming language. Using its vision capabilities to analyze overhead imagery, Claude devised a point-by-point path for Perseverance. The AI iterated on its work, critiquing and refining its own waypoint suggestions.
To ensure safety, Claude's proposed routes underwent rigorous simulation, modeling over 500,000 variables to predict rover positions and identify potential hazards. While minor adjustments were made based on ground-level imagery, Claude's plans were largely accurate. This successful execution demonstrates the potential for AI in enabling more efficient autonomous rover navigation.
The engineers estimate that integrating Claude into the planning process could cut route-planning time in half. This efficiency gain allows for more frequent drives, increased data collection, and deeper analysis, ultimately accelerating our understanding of Mars.
Future of AI in space
This achievement with the Claude AI Mars rover is a precursor to broader AI integration in space exploration. The AI's ability to understand novel situations, write operational code, and make decisions with minimal human oversight is crucial for ambitious missions like NASA's Artemis campaign to return humans to the Moon.
Looking further ahead, autonomous AI systems could enable probes to explore distant solar system bodies like Europa or Titan. These missions face extreme conditions, including prolonged signal delays and harsh environments, where AI decision-making will be paramount.
