Google Public Sector has built a proof-of-concept robotic system with NASA aimed at supporting astronauts on future deep-space missions, including the Artemis program's planned Moon Base and eventual crewed trips to Mars.
The project responds to a specific problem: a round trip radio signal between Mars and Earth can take up to 44 minutes, making real-time medical guidance from mission control nearly impossible.
NASA's Human Research Program, the Johnson Space Center Robotics Team and the Jet Propulsion Laboratory (JPL) worked with Google Public Sector on the system, which follows last year's Crew Medical Officer Digital Assistant, a clinical decision-support tool for remote medical guidance.
The robot is designed as a general-purpose "payload handler" that manages cargo and equipment during routine operations, then switches to medical support roles, such as retrieving and preparing an ultrasound machine, when needed.
JPL adapted satellite situational-awareness tools, originally built to track objects in orbit, to give the robot spatial awareness in tight spacecraft quarters.
Engineers tested the system at Johnson Space Center's iMETRO facility using Google DeepMind's MuJoCo physics simulator before running it on real hardware.
In a recent demonstration, the robot performed a "pick-and-place" task locating and moving a medical bag, achieving an 85% baseline success rate using an open-source vision-language-action model called Pi0, built by Physical Intelligence on Google Cloud infrastructure.
Because spacecraft cannot rely on continuous links to Earth-based data centers, Google fine-tuned the 4-billion-parameter Pi0 model on Google Cloud G4 instances in 2.5 hours.
Then ran the model locally on compact edge hardware; the team has tested around 15 edge devices over two years to find ones suited to low-power missions.
The eventual goal, Google Public Sector said, is a system astronauts can direct with simple voice commands for tasks such as capturing vitals or basic ultrasounds, under a human crew medical officer's supervision throughout.
