Progress-Think: Semantic Progress Reasoning for Vision-Language Navigation

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Wang, Shuo, Wang, Yucheng, Lian, Guoxin, Wang, Yongcai, Chen, Maiyue, Wang, Kaihui, Zhang, Bo, Su, Zhizhong, Zhou, Yutian, Li, Wanting, Li, Deying, Fan, Zhaoxin
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911592700968960
author Wang, Shuo
Wang, Yucheng
Lian, Guoxin
Wang, Yongcai
Chen, Maiyue
Wang, Kaihui
Zhang, Bo
Su, Zhizhong
Zhou, Yutian
Li, Wanting
Li, Deying
Fan, Zhaoxin
author_facet Wang, Shuo
Wang, Yucheng
Lian, Guoxin
Wang, Yongcai
Chen, Maiyue
Wang, Kaihui
Zhang, Bo
Su, Zhizhong
Zhou, Yutian
Li, Wanting
Li, Deying
Fan, Zhaoxin
contents Vision-Language Navigation requires agents to act coherently over long horizons by understanding not only local visual context but also how far they have advanced within a multi-step instruction. However, recent Vision-Language-Action models focus on direct action prediction and earlier progress methods predict numeric achievements; both overlook the monotonic co-progression property of the observation and instruction sequences. Building on this insight, Progress-Think introduces semantic progress reasoning, predicting instruction-style progress from visual observations to enable more accurate navigation. To achieve this without expensive annotations, we propose a three-stage framework. In the initial stage, Self-Aligned Progress Pretraining bootstraps a reasoning module via a novel differentiable alignment between visual history and instruction prefixes. Then, Progress-Guided Policy Pretraining injects learned progress states into the navigation context, guiding the policy toward consistent actions. Finally, Progress-Policy Co-Finetuning jointly optimizes both modules with tailored progress-aware reinforcement objectives. Experiments on R2R-CE and RxR-CE show state-of-the-art success and efficiency, demonstrating that semantic progress yields a more consistent representation of navigation advancement.
format Preprint
id arxiv_https___arxiv_org_abs_2511_17097
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Progress-Think: Semantic Progress Reasoning for Vision-Language Navigation
Wang, Shuo
Wang, Yucheng
Lian, Guoxin
Wang, Yongcai
Chen, Maiyue
Wang, Kaihui
Zhang, Bo
Su, Zhizhong
Zhou, Yutian
Li, Wanting
Li, Deying
Fan, Zhaoxin
Robotics
Vision-Language Navigation requires agents to act coherently over long horizons by understanding not only local visual context but also how far they have advanced within a multi-step instruction. However, recent Vision-Language-Action models focus on direct action prediction and earlier progress methods predict numeric achievements; both overlook the monotonic co-progression property of the observation and instruction sequences. Building on this insight, Progress-Think introduces semantic progress reasoning, predicting instruction-style progress from visual observations to enable more accurate navigation. To achieve this without expensive annotations, we propose a three-stage framework. In the initial stage, Self-Aligned Progress Pretraining bootstraps a reasoning module via a novel differentiable alignment between visual history and instruction prefixes. Then, Progress-Guided Policy Pretraining injects learned progress states into the navigation context, guiding the policy toward consistent actions. Finally, Progress-Policy Co-Finetuning jointly optimizes both modules with tailored progress-aware reinforcement objectives. Experiments on R2R-CE and RxR-CE show state-of-the-art success and efficiency, demonstrating that semantic progress yields a more consistent representation of navigation advancement.
title Progress-Think: Semantic Progress Reasoning for Vision-Language Navigation
topic Robotics
url https://arxiv.org/abs/2511.17097