LATIDIA · Robótica

GG-Nav: La cadena de pensamiento de puesta a tierra orientada a objetivos provoca razonamiento en los modelos de base de navegación

arXiv:2610.14541v1 Announce Type: new Abstract: Emerging strong Vision-Language Models (VLMs) have advanced navigation foundation models. Sin embargo, estos modelos todavía tienen un rendimiento insatisfactorio en la navegación orientada a objetivos

La noticia

arXiv:2610.14541v1 Announce Type: new Abstract: Emerging strong Vision-Language Models (VLMs) have advanced navigation foundation models. However, these models still have unsatisfactory performance on goal-oriented navigation settings. In this paper, we first conduct a motivating study revealing that the key issue lies in the insufficient utilization of streaming observations and imprecise target localization. To tackle these challenges, we propose GG-Nav, a simple yet effective navigation framework that learns a structured chain-of-thought (CoT) mechanism via supervised fine-tuning to explicitly localize goal-relevant instances, accumulate grounding evidence, and support more reliable decision-making. Specifically, the structured CoT is designed with two key stages: (i) Observation-to-Evidence, which reasons over all intermediate frames between adjacent decision steps