Grasp Affordance

这是一个用于视觉抓取可供性预测的数据集，它促进了更强大和异构的机器人抓取方法。该数据集包含来自 30 个不同对象的不同属性。每个对象实例不仅与语义描述相关，还与描述与各种动作相关的视觉属性、位置和不同抓取区域的物理特征相关。This is a dataset for visual grasp affordance prediction, which facilitates the development of more robust and heterogeneous robotic grasping approaches. This dataset encompasses diverse attributes across 30 distinct objects. Each object instance is associated with not only semantic descriptions, but also physical features that depict visual attributes, positions and distinct grasping regions relevant to various actions.

搜集汇总

背景与挑战

背景概述

该数据集专注于视觉抓取可供性预测，旨在提升机器人抓取方法的鲁棒性和多样性。它包含30个不同对象的实例数据，每个实例关联了语义描述、视觉属性、位置信息及抓取区域的物理特征。