NASA’s Jet Propulsion Laboratory asked Anthropic’s AI model Claude to plan a short drive for the Perseverance rover on Mars. The rover received the instructions on Dec 8 and Dec 10 2025 and successfully navigated a 400‑meter rocky path. Until now, human engineers have drawn every waypoint by hand.
How Claude helped the rover
Engineers fed Claude images of the terrain and years of driving data. Using its coding skills, Claude wrote commands in Rover Markup Language, the XML‑based language used for Mars rovers. It broke the route into ten‑meter segments, refined the waypoints, and ran a simulation that checks over 500,000 variables for hazards. JPL staff only made minor tweaks after reviewing the plan.
What this means for future missions
The team estimates that Claude halved the time needed to plan a drive. Faster, consistent planning lets operators schedule more drives and collect more science. NASA sees the same approach as useful for upcoming Moon missions and for probes that will travel far beyond Mars, where communication delays are even longer.
Why it matters
For the public, AI‑assisted planning means future rovers can explore more ground faster, delivering more scientific data without waiting for lengthy human input. It also shows how AI could help manage the long communication delays that will affect missions to the Moon, Europa or Titan.
