Space-Native AI 2026:
Intelligence Beyond Earth
An open call for research questions and project concepts on machine learning, autonomous systems and scientific discovery beyond Earth.
Submit to Cycle I ↗Space should change the research question, not just the deployment location.
Space-Native AI begins with a simple premise: unfamiliar physics, limited communication, scarce experiments, constrained computation and high autonomy can change the structure of the learning problem itself.
Cycle I invites researchers, engineers, students and interdisciplinary teams to contribute short, non-confidential research questions or project concepts that help define what this emerging field should investigate next.
The aim is not merely to collect ideas. After the call closes, contributions will be synthesized into themes, selected submissions will receive editorial feedback, community input will be invited on emerging directions, and a small number of Featured Contributions will be highlighted in the first Cycle Review.
Three core challenges.
Learning under altered physical conditions
Research angles include dynamics shifts, world modeling, simulation-to-reality transfer, continual adaptation and uncertainty calibration.
Closed-loop scientific discovery
Research angles include active learning, scientific agents, optimal experiment design, robotic laboratories and hypothesis refinement.
Intelligence under severe constraints
Research angles include constrained decision-making, information gain, data efficiency, reliability and risk-aware planning.
Submit → synthesize → discuss → connect.
The first cycle is designed to turn scattered ideas into a visible research landscape and a starting point for a durable community.
Research questions & project concepts
Participants submit a concise question or concept, explain why the space setting matters, and describe what meaningful progress could look like.
Theme building
The organizing team clusters contributions into emerging research areas and identifies recurring technical questions, missing capabilities and possible project directions.
Editorial feedback and community discussion
Selected submissions receive concise feedback and may be opened for public comments or invited into topic-focused follow-up discussions.
Featured Contributions and next steps
A small number of especially generative contributions will be highlighted in the Cycle I Review. Where useful, contributors working on related problems may be connected for future community activities or collaborations.
One Liter in Space
What scientific problem deserves one liter?
Imagine a conceptual 10 × 10 × 10 cm experimental volume. What should it investigate, what should it observe, and how could an intelligent system decide what to do next when every experiment and resource counts?
This is a thought-experiment boundary, not an offer of flight hardware, launch capacity or a flight-ready CubeSat specification.
Submit a One Liter concept ↗Recognition with a purpose.
Featured Contributions are intended to surface questions that can generate useful research and bring related people into the same conversation.
Selection will consider research significance, whether the space environment genuinely changes the problem, relevance to AI or learning, and the contribution's potential to lead to discussion, collaboration, a benchmark, an experiment or a future project.
Featured status is community recognition and editorial selection. It is not peer review, publication acceptance, funding, or a scientific endorsement.
Submit a question. Help map the field. Find the people who care about the same problem.
Space-Native AI is independently organized and coordinated from Darmstadt, Germany. Participation is free and open internationally.
Submit to Cycle I ↗Space-Native AI does not imply endorsement by World Space Week, the United Nations, ESA, TU Darmstadt or any academic conference.