Google and NASA Test AI Robot for Deep-Space Missions

The robotic system successfully managed logistics using an advanced vision-language-action model during recent laboratory tests.

Updated on Oct. 28, 2026 in Robotics

Isometric editorial illustration of a high-tech modular robotic arm unit, symbolizing autonomous systems for deep-space missions. — Google Cloud and NASA are evaluating an AI-driven robotic system capable of managing mission logistics independently to mitigate communication delays for future deep-space travel. AI Illustration. Upload story photo >

Live Poll

Do you trust AI-powered robotic systems to perform critical tasks without direct human supervision?

Google Cloud and NASA recently tested an AI-powered robotic system designed to assist with logistics and medical support for deep-space missions. The platform uses a vision-language-action model to help astronauts handle mission-critical work while managing tasks independently.

Why it matters

This technology aims to decrease reliance on Earth-based computing, which is essential for Mars missions that face communication delays of up to 44 minutes. By processing data locally, the system allows astronauts to perform complex tasks without real-time guidance from ground control.

The system utilizes a Pi0 model containing 4 billion parameters that was fine-tuned in approximately 2.5 hours. Researchers also evaluated 15 edge-computing devices to support local AI processing capabilities in space hardware.

The players

NASA

The United States government agency responsible for the nation's civilian space program and for aeronautics and space research.

Google Cloud

A suite of cloud computing services offered by Google that provides infrastructure and machine learning tools.

Jet Propulsion Laboratory

A federally funded research and development center managed by Caltech for NASA that specializes in robotic exploration.

Johnson Space Center

The home of NASA's human spaceflight training programs and mission control for space operations.

The details

The robotic system demonstrated the ability to locate and transport a medical bag during trials at the Johnson Space Center. Engineers utilized Google DeepMind's MuJoCo physics simulator for training, while the Jet Propulsion Laboratory provided orbital object-tracking technology to improve the robot's spatial awareness.

From 2024 to 2026, the team evaluated approximately 15 edge-computing devices.
Testing for the AI-powered robot system took place in October 2026.

The Tech Race

This project builds directly on the Crew Medical Officer Digital Assistant system to expand autonomous capabilities in hazardous environments. The development signifies a shift toward decentralized space computing as agencies race to overcome the limitations of distance in long-duration missions.

The system could eventually allow astronauts to issue commands using spoken or written language rather than manual controls. By automating supply management, the technology simplifies the daily workload for mission crews during long-duration flight.

The takeaway

Advancements in local AI processing are becoming the backbone of future deep-space exploration. As these systems evolve, they demonstrate how edge computing can solve the critical problem of communication latency in remote environments.