发表机构
Institute of Robotics and Machine Intelligence, Poznan University of Technology; CSIRO Robotics, CSIRO(波兹南理工大学机器人与机器智能研究所; 联邦科学与工业研究组织机器人部)
机构由 AI 辅助整理,请以论文原文为准。AI 中文总结
本文提出NEO框架,用于机器人操纵的语言引导NeRF编辑。结合神经场重采样与多视图修复实现物体移除,利用知识蒸馏进行NeRF权重编辑,还创建了评估基准NEO-Dataset。该方法在场景编辑任务中表现出色,优于现有基线。
AI 中文摘要
在本文中,我们提出了NEO,这是一个为机器人操纵提供语言引导的NeRF编辑的统一框架。我们的论文介绍了:(i)一种将神经场重采样与多视图一致的渐进式修复相结合的语言引导物体移除方法;(ii)一种利用知识蒸馏的直接NeRF权重编辑方法,通过师生模型组合原始和编辑后的NeRF,在机器人执行动作之前对未来场景状态进行连贯建模;(iii)第一个用于定量评估适用于机器人操纵的NeRF场景编辑方法的基准(NEO数据集)。我们表明,我们的方法在场景编辑任务中优于现有基线,包括物体移除和机器人抓取放置实验,产生视觉上连贯且几何上一致的编辑,减少了先前方法通常引入的伪影。
英文摘要
In this paper, we present NEO, a unified framework providing language-guided NeRF editing for robotic manipulation. Our paper introduces (i) a language-guided object removal that combines neural field resampling with multiview-consistent progressive inpainting, (ii) a direct NeRF weight editing method utilizing knowledge distillation, composing original and edited NeRFs via a teacher-student model, enabling coherent modeling of future scene states before a robot executes an action, and (iii) the first benchmark (NEO-Dataset) for quantitatively evaluating NeRF scene editing methods suitable for robot manipulation. We show that our approach outperforms state-of-the-art baselines in scene editing tasks, including object removal and pick-and-place robotic experiments, yielding visually coherent and geometrically consistent edits that reduce artifacts commonly introduced by prior methods.
CommentsAccepted to IEEE ROBOTICS AND AUTOMATION LETTERS (RA-L) JULY, 2026