UK Space Agency sets out a ten-year roadmap for trusted AI in space

The UK Space Agency set out a roadmap for AI development and adoption in space.

The UK Space Agency published an AI roadmap for space on October 27. It sets out a ten-year direction and draws on consultation with more than 100 stakeholders. The report identifies eight technology priorities and seven investment themes, with a focus on systems the UK could make reliable, certifiable and efficient enough for space use. [UK Space Agency roadmap](https://www.gov.uk/government/publications/artificial-intelligence-roadmap-for-space/artificial-intelligence-roadmap-for-space)

Onboard operations and analysis on Earth have different constraints

AI can help analyze Earth observation data on the ground or support a decision onboard a spacecraft. Those uses have different failure modes. A ground-based analyst can often rerun a model with more compute or inspect the input. Onboard software may have limited power, bandwidth and opportunities for human intervention. An operational team needs to know what happens when the model sees unfamiliar conditions or returns a result outside its useful range.

The roadmap connects applications such as Earth observation, resilient satellite communications and spacecraft autonomy with capabilities including assurance, skills and test infrastructure. Its argument is that the UK can compete through specialized systems and services. The document sets priorities and recommendations. It does not itself commission a mission or create a funded procurement. [UK Space Agency roadmap](https://www.gov.uk/government/publications/artificial-intelligence-roadmap-for-space/artificial-intelligence-roadmap-for-space)

From a promising demonstration to a mission decision

A promising ground demonstration may show that a model can recognize an event or help plan an operation. Space use raises further questions about test coverage, hardware limits and how operators can recover if the system behaves unexpectedly. Those requirements are especially important when software is expected to act with limited contact from Earth.

The roadmap gives researchers and suppliers a way to see which capability gaps the agency considers important. Future evidence of progress would include shared testing facilities, clear assurance requirements and demonstrations whose results can be traced to an operational decision. The October publication sets the direction for that work. [UK Space Agency roadmap](https://www.gov.uk/government/publications/artificial-intelligence-roadmap-for-space/artificial-intelligence-roadmap-for-space)