Abstract
The growing complexity of space missions has placed increasingly stringent demands on the autonomy and reliability of spacecraft. Advanced artificial intelligence (AI) technologies—particularly large language models (LLMs)—offer significant advantages in feature extraction, logical reasoning, and autonomous decision-making, owing to their massive parameter capacity, multimodal integration capabilities, and learning adaptability. These strengths have led to the expanding application of AI in the spacecraft domain, significantly enhancing system performance and mission efficiency. This paper systematically reviews the current status of AI-enabled spacecraft applications in major fields such as deep space exploration, remote sensing, communications, and navigation. A hierarchical “4+n” architecture is proposed to define the AI-driven spacecraft application system, comprising four core layers: infrastructure, platform, algorithm, and application. The development trend of AI in this domain is moving toward human-machine hybrid intelligence, swarm intelligence, and artificial general intelligence. Based on this perspective, four key research areas are identified: intelligent perception recognition and in-orbit data mining information fusion techniques; adaptive and self-evolving decision-making and control technologies for spacecraft; generalized space-ground learning for cooperative and game-theoretic autonomous satellite swarms; and lightweight, domain-specific large model technologies for in-orbit applications. These research directions provide theoretical foundations and technical references for the future advancement of AI in spacecraft systems.
| Translated title of the contribution | AI-enabled spacecraft applications: current research and future prospects |
|---|---|
| Original language | Chinese (Traditional) |
| Pages (from-to) | 2524-2541 |
| Number of pages | 18 |
| Journal | Scientia Sinica Informationis |
| Volume | 55 |
| Issue number | 10 |
| DOIs | |
| Publication status | Published - 1 Oct 2025 |
| Externally published | Yes |
Fingerprint
Dive into the research topics of 'AI-enabled spacecraft applications: current research and future prospects'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver