EUVIMEDEUVIMED entdecken
Crossref

Mathematically-Grounded, Explainable Multi-Agent Systems for Terrain Hyperspectral Analysis and Triage

Nasrulla K, Sakthivel C R

International Journal on Science and Technology · 15.07.2026 · 17 (3)

DOI: 10.71097/ijsat.v17.i3.11399 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal on Science and Technology, online erschienen am 15.07.2026. Verfasst von Nasrulla K und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Agentic systems, in which massive language models manage analytical tools and large foundation models that are trained on data from earth observation systems are two concurrent trends that are transforming the field of remote sensing. A literature review of 2025-2026 shows that both of them have more powerful and independent analysis but neither of them can meet the needs of high-stakes terrain analysis. The agentic remote-sensing systems of today are characterized by high tool-orchestration error, shallow non-replayable memory, hyperspectral and spectral-physics integration is an open challenge and reasoning about red-green-blue and multispectral tools. Foundation models generate strong embeddings but do not have statistical guarantees and decision provenance. While optimal transport is already a widely used and established mathematical tool for solving the problem, there is also rapid development of other tools such as tensor and block-term decomposition, topological data analysis, graph diffusion and conformal prediction, which are each emerging as a stand-alone tool rather than as a common analytical substrate over which an agent can reason. In this survey, 20 representative papers from the year 2025-2026 related to geospatial and agentic artificial intelligence, Earth foundation models, hyperspectral unmixing and classification, topological and optimal-transport techniques, uncertainty quantification, explainable AI and change and tamper detection are discussed. It organizes them by the research topic that the theoretically inspired, explainable, LLM-independent multi-agent framework TerraXIA tackles. Unlike in other systems, each agent in TerraXIA generates uncertainty-bounded, verifiable triage based on a common mathematical representation of the world-state of a terrain hyperspectral scene. The review provides a consensus scientific basis for the recommended work program by identifying a consistent set of gaps and the relationship of each gap to the four research objectives and 12 innovations in the framework

Zitieren

APA

K, N., & R, S. C. (2026). Mathematically-Grounded, Explainable Multi-Agent Systems for Terrain Hyperspectral Analysis and Triage. International Journal on Science and Technology, 17(3). https://doi.org/10.71097/ijsat.v17.i3.11399

Vancouver

K N, R SC. Mathematically-Grounded, Explainable Multi-Agent Systems for Terrain Hyperspectral Analysis and Triage. International Journal on Science and Technology. 2026;17(3). doi:10.71097/ijsat.v17.i3.11399

BibTeX
@article{K2026,
  author = {K, Nasrulla and R, Sakthivel C},
  title = {Mathematically-Grounded, Explainable Multi-Agent Systems for Terrain Hyperspectral Analysis and Triage},
  journal = {International Journal on Science and Technology},
  year = {2026},
  volume = {17},
  number = {3},
  doi = {10.71097/ijsat.v17.i3.11399}
}
Kostenlos starten →