NASA used Claude to plot a route for its Perseverance rover on Mars

Since 2021, NASA’s Perseverance rover has achieved several historic milestones, including sending back the first audio recording from Mars. Now, nearly five years after landing on the Red Planet, it has achieved another feat. Last December, Perseverance successfully completed a route through a section of Jezero Crater plotted by Anthropic’s cloud chatbot, marking the first time that NASA has used a large language model to drive a car-sized robot.

Between December 8 and 10, Perseverance traveled about 400 meters (about 437 yards) through a field of rocks on the surface of Mars mapped by the cloud. As you can imagine, using an AI model to plot a course for persistence wasn’t as simple as inputting a signal.

As NASA points out, crossing Perseverance is no easy task, even for a human. “Each rover drive must be carefully planned so that the machine does not slip, tip, spin, or beach itself,” NASA said. “So from the moment the rover landed, its human operators, using a combination of images taken from space and the rover’s onboard cameras, have painstakingly determined waypoints for it to follow – they call it the ‘breadcrumb trail’.”

To get the cloud to complete the task, NASA had to first provide Anthropic’s programming agent cloud code with “years” of relevant data from the rover before the model could begin writing a path to Perseverance. Cloud then systematically proceeded the mapping process, piecing together waypoints from ten-meter segments that were then reviewed and repeated.

This is NASA we’re talking about, and engineers at the agency’s Jet Propulsion Laboratory (JPL) made sure to double-check the model’s work before sending it off to Perseverance. The JPL team ran Cloud’s waypoints through simulations, which they use every day to confirm the accuracy of the commands they send to the rover. In the end, NASA says it had to make only “minor changes” to Cloud’s route, a change that resulted from the fact that the team had access to ground-level images that Cloud had not seen in its planning process.

“Engineers estimate that using the cloud in this way will cut route-planning time in half and make trips more consistent,” NASA said. “Spending less time doing arduous manual planning – and spending less time training – allows the rover’s operators to fit in even more drives, collect even more scientific data, and conduct even more analysis. This means, in short, that we’ll learn a lot more about Mars.”

While the productivity benefits provided by AI are often overstated, in the case of NASA, any tool that can make its scientists more efficient will certainly be welcomed. Over the summer, the agency lost about 4,000 employees — about 20 percent of its workforce — due to Trump administration cuts. Going into 2026, the President had proposed cutting the agency’s science budget by nearly half, before Congress ultimately rejected that plan in early January. Still, despite preserving its funding just below 2025 levels, the agency faces a tough road ahead. It is being asked to return to the Moon with less than half the workforce during the peak of the Apollo program.

Meanwhile, for Anthropic, this is a major achievement. You might remember that last spring even Claude couldn’t beat pokemon red. In less than a year, the company’s models have gone from struggling to navigate a simple 8-bit Game Boy game to successfully plotting a course for a rover on a distant planet. NASA is excited about the possibility of future collaboration, saying, “Autonomous AI systems could help probes explore more distant parts of the solar system.”