Multimodal-Human-Robot-Teaching-Dataset

Summary: This dataset was collected as part of the CISC (Collaborative Intelligence for Safety-Critical Systems) project under the Marie Skłodowska-Curie ITN programme. The data supports research in human-machine interaction and programming by demonstration in safety-critical environments.

It contains multimodal recordings, including eye-tracking data, human posture information (captured via a vision-based system), robot trajectory data, and questionnaire responses related to usability and cognitive load.

The dataset was collected through experiments involving human participants performing teaching tasks with a collaborative robot. All data has been anonymized in line with ethical approvals and is suitable for analysis in human factors, ergonomics, computer vision, and AI feedback systems.

This dataset can be used for academic research in machine learning, human-robot collaboration, and system evaluation studies.

Please cite this dataset if used in publications. Funding for this work was provided by the EU Horizon 2020 programme under Grant Agreement No.955901.

Multimodal Human-Robot Teaching Dataset: This dataset captures multimodal interaction data collected during a study on teaching robots through human demonstration. The data includes eye tracking, motion tracking, robot trajectory, and subjective evaluations from participants performing structured teaching tasks under different feedback conditions.

Experiment Overview: Participants(N=28) were asked to perform Pick-and-Place and Sliding tasks using a teaching-by-demonstration interface in a controlled experimental setup. Each participant experienced one of four feedback conditions: No Feedback-NF (Control) Visual Feedback-VF Descrptive Feedback-DF Mixed Feedback-MF Data was collected across multiple modalities to investigate how humans learn and teach robots, with an emphasis on usability, attention, and ergonomic performance.

participants.csv: Participant demographics (age, gender) subjective_responses.csv: NASA-TLX and SUS questionnaire responses per condition and task EyeTracking/: Gaze data and fixation metrics (participant → condition → task) MotionTracking/: Motion capture data

Use Case: This dataset can support research in: Human-robot interaction Cognitive load estimation Programming-by-demonstration systems Eye tracking analysis Multimodal machine learning

License: This dataset is shared under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

Citation If you use this dataset, please cite our related data paper (DOI to be added soon).

数据集概述：本数据集是欧盟玛丽·居里行动计划（Marie Skłodowska-Curie ITN）下CISC（Collaborative Intelligence for Safety-Critical Systems，安全关键系统协作智能）项目的研究采集数据，可用于安全关键环境下的人机交互与示教编程（Programming-by-Demonstration）研究。
该数据集包含多模态记录数据，涵盖眼动追踪（eye-tracking）数据、基于视觉系统采集的人体姿态信息、机器人轨迹数据，以及与可用性和认知负荷相关的问卷反馈。
本数据集通过人类参与者与协作机器人完成示教任务的实验采集所得，所有数据均已按照伦理审查要求完成匿名化处理，适用于人为因素、工效学、计算机视觉与AI反馈系统等领域的分析研究。
该数据集可用于机器学习、人机协作与系统评估等方向的学术研究。若在出版物中使用本数据集，请务必引用本数据集。本研究的资助方为欧盟地平线2020计划，资助协议编号为955901。

多模态人机示教数据集：本数据集采集了人类通过示教方式教导机器人的研究中产生的多模态交互数据，包含不同反馈条件下参与者完成结构化示教任务时的眼动追踪、动作追踪、机器人轨迹数据，以及主观评价结果。

实验概况：共招募28名参与者（N=28），在受控实验环境下通过示教编程界面完成拾取与放置（Pick-and-Place）及滑动（Sliding）两类任务。每名参与者需接受四种反馈条件之一：无反馈（NF，对照组）、视觉反馈（VF）、描述性反馈（DF）以及混合反馈（MF）。本数据集通过多模态数据采集，旨在探究人类如何学习与教导机器人，重点关注可用性、注意力分配与工效学表现。

数据文件说明：
participants.csv：存储参与者人口统计学信息（年龄、性别）
subjective_responses.csv：存储各条件与任务下的NASA任务负荷指数（NASA-TLX）与系统可用性量表（SUS）问卷反馈
EyeTracking/：存储注视数据与注视指标（按参与者→条件→任务分级存储）
MotionTracking/：存储动作捕捉数据

应用场景：本数据集可支撑以下方向的研究：人机交互、认知负荷评估、示教编程系统、眼动追踪分析、多模态机器学习。

许可协议：本数据集采用知识共享署名4.0国际（Creative Commons Attribution 4.0 International，CC BY 4.0）许可协议进行共享。

引用说明：若在研究中使用本数据集，请引用本数据集相关的学术论文（DOI信息将在后续补充）。