NASA Guided Perseverance Rover on Mars Using Artificial Intelligence

AI Plots Course on Martian Surface

NASA's Mars rover Perseverance has undertaken a new technological trial in the fifth year of its mission. According to a statement from the Jet Propulsion Laboratory (JPL), during two separate drives on Martian days 1707 and 1709, the rover followed a route planned by Claude, a generative AI model developed by the California-based company Anthropic. The rover traveled 210 and 246 meters respectively on these drives.

The Challenging Process Stretching from Earth to Mars

The average distance of 225 million kilometers between Earth and Mars causes significant communication delays. This delay prevents the rover from reacting instantly to unexpected obstacles on the Martian surface. Past examples, such as the Spirit rover getting stuck in a sand trap and ending its mission, demonstrate how critical route planning is. Using traditional methods, engineers meticulously analyze images from space and the rover to create routes that avoid obstacles like rocks, outcrops, and dunes.

How Did the AI Planning Work?

Prior to the trial, JPL engineers fed data to the Claude AI model on how to operate the rover. The commands generated by the AI were validated through simulations controlling over 500,000 variables. According to Anthropic's statement, Claude's commands were found to be surprisingly smooth, requiring only minor adjustments for some images the model hadn't seen before. After the team was satisfied with the results, the plan written by the AI was transmitted to Perseverance in December and successfully executed.

Potential to Double Exploration Speed

Engineer Vandi Verma from the JPL Perseverance team stated that generative AI shows great potential in facilitating the building blocks of autonomous navigation for extraterrestrial driving. Anthropic expressed that this program could cut route planning time in half and allow for much more information to be gathered about Mars. Verma predicted that although Perseverance has only traveled a few hundred meters for now, AI could soon plan routes kilometers long on the Martian surface.

NASA Administrator Jared Isaacman emphasized that this demonstration showcases how much capabilities have advanced and expands how they will explore other worlds. Isaacman noted that this is a powerful example of teams carefully and responsibly implementing new technology in real operations.

This development is seen as a sign that AI's role in space exploration will increasingly grow. In future missions, the widespread use of similar autonomous systems is expected. Rapid advancements in the field of AI continue to deeply affect company strategies as well.