MultiPhysio-HRC: Multimodal Physiological Signals Dataset for industrial Human-Robot Collaboration

MultiPhysio-HRC is a multimodal dataset collected to study mental state perception in industrial Human–Robot Collaboration (HRC) scenarios. The dataset includes synchronized physiological, audio, and facial data acquired during controlled cognitive tasks, immersive virtual reality experiences, and real-world industrial disassembly tasks performed both manually and in collaboration with a robot. Recorded modalities include EEG, ECG, electrodermal activity (EDA), respiration (RESP), electromyography (EMG), together with voice recordings and facial action units. Ground-truth annotations were obtained using validated self-assessment questionnaires, including STAI-Y1, NASA-TLX, SAM, and NARS, enabling the study of stress, cognitive load, and affective states. MultiPhysio-HRC is designed to support research in affective computing, multimodal learning, and human-aware robotics, and to foster the development of adaptive robotic systems aligned with the human-centric vision of Industry 5.0.The dataset documentation, structure, and feature descriptions are provided in the accompanying README.MultiPhysio-HRC是一款用于研究工业人机协作（Human–Robot Collaboration, HRC）场景中心理状态感知的多模态数据集。该数据集涵盖了受控认知任务、沉浸式虚拟现实体验，以及人工独立完成或与机器人协作开展的真实工业拆卸任务中采集的同步生理、音频与面部数据。所采集的模态包括脑电图（EEG）、心电图（ECG）、皮肤电活动（electrodermal activity, EDA）、呼吸（respiration, RESP）、肌电图（electromyography, EMG），同时还包含语音录音与面部动作单元数据。该数据集的真值标注通过经过信效度验证的自我评估问卷获取，包括状态-特质焦虑量表Y1分量表（STAI-Y1）、NASA任务负荷指数量表（NASA-TLX）、主观情感评价量表（SAM）与负面情感反应量表（NARS），可用于开展压力、认知负荷与情感状态的相关研究。MultiPhysio-HRC旨在为情感计算、多模态学习与人性化机器人学领域的研究提供支撑，并助力适配以人为本的工业5.0（Industry 5.0）愿景下的自适应机器人系统的开发。相关数据集文档、数据结构与特征描述详见配套的README文件。