Generative AI Gives Spacecraft the Autonomy Engineers Once Feared

Space engineering is moving toward greater use of generative AI and autonomous systems for spacecraft, though the technology remains far from trustworthy enough to hand over full control, according to IEEE Spectrum.

Last December, NASA's Jet Propulsion Laboratory used Anthropic's Claude models to help plan two Mars drives for the Perseverance rover, with human planners checking and adjusting the route before upload. In May, NASA and IBM placed a compressed AI model on the International Space Station and a satellite to identify floods and clouds from orbit, described as the first model of its kind demonstrated in space. In July, astronauts on the ISS tested a large language model to see whether it could help answer questions on maintenance procedures.

Robert Ambrose, former chief of NASA's Software, Robotics, and Simulation Division, said autonomy often lacks a deterministic outcome, meaning how a system arrived at a situation can change its behavior. He cited work on the Orion spacecraft and Robonaut 2, which went to space in 2011, and said engineers managed testing complexity partly by automating testing itself. He pointed to a possible Europa mission diving through a water plume as an example where Earth-based direction would come too late.

Icarus Robotics is developing a free-flying robotic system called Joy, which recently completed zero-gravity testing in Canada ahead of a planned ISS deployment, with an early task of moving cargo bags between modules. Cofounder and CTO Jamie Palmer said rollout would likely begin with partial autonomy and human supervision, noting that robots trained on Earth physics would fail in zero-g. Cofounder and CEO Ethan Barajas said no meaningful real-world dataset for that environment exists today.

Ufuk Topcu, an engineering professor at the University of Texas at Austin and director of the Center for Autonomy, said the goal cannot be guaranteeing autonomous systems never err, arguing instead for starting with restricted applications and gradually expanding use. He noted the space industry's shift from slow government mission cycles to faster commercial development.