More than A Point: Capturing Uncertainty with Adaptive Affordance Heatmaps for Spatial Grounding in Robotic Tasks

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Shao, Xinyu, Tang, Yanzhe, Xie, Pengwei, Zhou, Kaiwen, Zhuang, Yuzheng, Quan, Xingyue, Hao, Jianye, Zeng, Long, Li, Xiu
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908595191283712
author Shao, Xinyu
Tang, Yanzhe
Xie, Pengwei
Zhou, Kaiwen
Zhuang, Yuzheng
Quan, Xingyue
Hao, Jianye
Zeng, Long
Li, Xiu
author_facet Shao, Xinyu
Tang, Yanzhe
Xie, Pengwei
Zhou, Kaiwen
Zhuang, Yuzheng
Quan, Xingyue
Hao, Jianye
Zeng, Long
Li, Xiu
contents Many language-guided robotic systems rely on collapsing spatial reasoning into discrete points, making them brittle to perceptual noise and semantic ambiguity. To address this challenge, we propose RoboMAP, a framework that represents spatial targets as continuous, adaptive affordance heatmaps. This dense representation captures the uncertainty in spatial grounding and provides richer information for downstream policies, thereby significantly enhancing task success and interpretability. RoboMAP surpasses the previous state-of-the-art on a majority of grounding benchmarks with up to a 50x speed improvement, and achieves an 82\% success rate in real-world manipulation. Across extensive simulated and physical experiments, it demonstrates robust performance and shows strong zero-shot generalization to navigation. More details and videos can be found at https://robo-map.github.io.
format Preprint
id arxiv_https___arxiv_org_abs_2510_10912
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle More than A Point: Capturing Uncertainty with Adaptive Affordance Heatmaps for Spatial Grounding in Robotic Tasks
Shao, Xinyu
Tang, Yanzhe
Xie, Pengwei
Zhou, Kaiwen
Zhuang, Yuzheng
Quan, Xingyue
Hao, Jianye
Zeng, Long
Li, Xiu
Robotics
Many language-guided robotic systems rely on collapsing spatial reasoning into discrete points, making them brittle to perceptual noise and semantic ambiguity. To address this challenge, we propose RoboMAP, a framework that represents spatial targets as continuous, adaptive affordance heatmaps. This dense representation captures the uncertainty in spatial grounding and provides richer information for downstream policies, thereby significantly enhancing task success and interpretability. RoboMAP surpasses the previous state-of-the-art on a majority of grounding benchmarks with up to a 50x speed improvement, and achieves an 82\% success rate in real-world manipulation. Across extensive simulated and physical experiments, it demonstrates robust performance and shows strong zero-shot generalization to navigation. More details and videos can be found at https://robo-map.github.io.
title More than A Point: Capturing Uncertainty with Adaptive Affordance Heatmaps for Spatial Grounding in Robotic Tasks
topic Robotics
url https://arxiv.org/abs/2510.10912