Artificial intelligence (AI) is expected to think along—even in space. But conventional computers quickly reach their limits in orbit. A new study involving ECDF professor Janik Wolters (TU Berlin) presents an innovative approach: a computing system that operates not with electric current, but with light. The publication, titled “Concept of an in-orbit AI-system based on optical computing”, was developed within the project OMLO – Optical Computing for Machine Learning in Orbit. Contributors include Felix Kübler, Mingwei Yang, Lennart Mannteuffel, Okan Akyüz, Janik Wolters, and Enrico Stoll.

Whether in satellites or Mars rovers, AI can help analyze data directly on-site—quickly, efficiently, and autonomously. But powerful AI systems require a lot of energy, generate heat, and are vulnerable to radiation. Especially on small satellites, space is limited and power budgets are tight. The solution: Optical computing. “Key computing steps—such as image recognition—are carried out using light signals. This saves energy, is resistant to the harsh conditions of space, and allows for exceptionally fast calculations,” explains Janik Wolters, ECDF Professor for Physical Foundations of IT Security, about the new concept.

The study outlines a first concept for such a system—compact, flexible in use, and with promising potential for future applications. While the prototype is still experimental and not yet space-ready, initial tests show potential. Particularly exciting: the computing power of such an optical system increases disproportionately with the amount of data—while energy consumption remains relatively low. This makes it ideal for AI applications that process large volumes of data, such as satellite image analysis.

Next Steps: From Idea to Application

The researchers are already planning the next steps: miniaturization, tests under real-world conditions—and the search for research and industry partners to further develop the system. The goal: AI systems that can learn and make decisions autonomously in space—with the help of light.

To the publication (Open Access): //here