South SFV Today

A new center at the Stanford School of Engineering will leverage artificial intelligence in the service of space science, exploration, and business. The Center for AEroSpace Autonomy Research (CAESAR) aims to make these activities more efficient, safe, and sustainable.

Researchers at the center believe that AI could optimize navigation for spacecraft; land space vehicles on planets or asteroids; allow unmanned rovers to make autonomous decisions; monitor space debris around Earth; and address other challenges.

Marco Pavone, CAESAR co-founder and associate professor of aeronautics and astronautics, announced that one of the center’s main projects is developing a foundation model for space pursuits. This "space foundation model" will synthesize information across various modalities such as vision, text, remote sensing, and space-object catalogs. It will be capable of addressing tasks including situational awareness, positioning, and navigation.

"We want to develop rigorous tools for the trusted deployment of AI for spacecraft systems – trusted in the sense that they can behave within bounds described by the user," said Simone D’Amico, CAESAR co-founder and associate professor of aeronautics and astronautics.

D’Amico emphasized caution in incorporating AI into spacecraft systems due to potential pitfalls. He noted that sometimes AI components might not be the most effective choice for space systems.

"We founded CAESAR with the objective to tackle unsolved problems in spaceborne autonomy through the judicious incorporation of artificial intelligence components," D’Amico added.

Challenges include training AI in harsh environments like space and ensuring microprocessors' resilience against radiation. Pavone and D’Amico spoke during a symposium marking CAESAR's official launch on May 22. The event featured presentations from Stanford faculty members, postdoctoral scholars, students, NASA representatives, Aerospace Corp., Redwire Space, Blue Origin, and Lockheed Martin.

CAESAR's initial focus has been on developing machine learning models for maneuvers such as space rendezvous and docking. One such model is the Spacecraft Pose Estimation Network (SPN), which integrates machine learning with classical navigation algorithms to estimate a target spacecraft’s position from monocular images. Another model is the Autonomous Rendezvous Transformer (ART), designed to optimize spacecraft trajectories using a combination of AI-generated initial guesses refined by traditional mathematical optimization methods.

In another area of research at CAESAR is ReachBot—an autonomous robot equipped with extendable booms intended for Martian terrain exploration. Daniel Morton, a graduate student in mechanical engineering involved with ReachBot’s development explained its capabilities: "ReachBot can descend into a Martian lava tube...and drill into the wall to extract material for analysis."

The kickoff event also highlighted faculty unaffiliated with CAESAR but interested in related projects. Manan Arya discussed anchoring a radio reflector on the moon's far side to study early universe signals shielded from Earth's radio interference.

Grace Gao presented her lab's work on creating lunar navigation systems similar to GPS using small satellites orbiting the moon: "We want smaller satellites—as small as a shoebox." Her team collaborates with NASA’s Jet Propulsion Laboratory on CADRE—a project aiming to deploy three autonomous rovers on the moon next year.

Pavone expressed hope that collaboration would enhance CAESAR’s research impact: “We look forward to engaging with many people here today and the broader space community.”

ReachBot is jointly developed by Marco Pavone’s lab along with Mark Cutkosky's lab at Stanford University School of Engineering and Mathieu Lapôtre's lab from Earth Sciences department. Redwire Space and Blue Origin are co-sponsors of CAESAR.