LATIDIA · Robótica

Estrategias espaciales, no acciones: geodésica cuantificada por vectores como herramientas para agentes impulsados por LLM

arXiv: 2610.00613v1Announce Type: cross Resumen: Los agentes basados en el modelo de lenguaje grande (LLM) a menudo son criticados por carecer de comprensión espacial y explotar principalmente patrones de texto estadísticos. Investigamos su espacialidad

La noticia

arXiv:2610.00613v1 Announce Type: cross Abstract: Large language model (LLM) based agents are often criticized for lacking spatial understanding and mainly exploiting statistical text patterns. We investigate their spatial comprehension through an architecture combining geometrical tools with a LLM serving as a high-level orchestrator in grid-world environments. The agent first collects geodesic trajectories, which are then vector-quantized to extract a representative subset. Offline, the LLM associates a natural language description of the underlying behavioral patterns to each selected trajectory, making it a tool. Online, the LLM chooses the appropriate tool conditioned on the current state and goal. Low-level control is handled by primitive actions that execute the trajectory associated with the