HRIBench
收藏资源简介:
HRIBench是一个视觉问答(VQA)基准,旨在评估视觉语言模型(VLMs)在人类-机器人交互(HRI)中感知人类行为的能力。该数据集包含五个关键领域:非言语提示理解、言语指令理解、人-机器人-物体关系理解、社交导航和个人识别。HRIBench通过从真实世界的HRI环境中收集数据,并对剩余四个领域利用公开可用的数据集来构建。每个领域整理了200个VQA问题,总共1000个问题。该数据集的应用领域旨在解决实时HRI中的核心感知能力问题,例如理解细粒度多模态提示、解决模糊的语言-视觉指令和进行现实世界的空间和物理推理。
HRIBench is a visual question answering (VQA) benchmark designed to evaluate the capability of vision-language models (VLMs) to perceive human behaviors in human-robot interaction (HRI) scenarios. This dataset covers five core domains: nonverbal cue comprehension, verbal instruction understanding, human-robot-object relationship comprehension, social navigation, and person identification. HRIBench is constructed by collecting data from real-world HRI environments and leveraging publicly available datasets for the remaining four domains. Each domain contains 200 curated VQA questions, totaling 1,000 questions overall. The target application scenarios of this dataset aim to address core perceptual capabilities in real-time HRI, such as understanding fine-grained multimodal prompts, resolving ambiguous language-visual instructions, and conducting real-world spatial and physical reasoning.
HRIBench数据集概述
数据集简介
- HRIBench是一个视觉问答(VQA)基准测试数据集,专为评估视觉语言模型(VLMs)在多种人类感知任务上的表现而设计。
数据集目的
- 旨在评估视觉语言模型在与人机交互(HRI)相关的关键人类感知任务中的性能。
适用领域
- 视觉问答(VQA)
- 人机交互(HRI)
- 视觉语言模型(VLMs)评估




