Dataset: Video recordings of human-robot interactions with a Nao robot controlled via the SONAR adaptive control architecture for social norm aware robots

# GENERAL INFORMATION<br>## TitleVideo recordings of human-robot interactions with a Nao robot controlled via the SONAR adaptive control architecture for social norm aware robots   ### Dataset DOI[https://doi.org/10.4121/50c7a19c-fc0e-4ef3-b35a-dd23bf08470d](https://doi.org/10.4121/50c7a19c-fc0e-4ef3-b35a-dd23bf08470d)<br>## Authors: - **Dr. Davide Dell'Anna** ([ORCID](https://orcid.org/0000-0002-1162-8341), [e-mail], Utrecht University, The Netherlands, [Personal webpage](https://www.davidedellanna.com) - **Dr. Anahita Jamshidnejad** ([ORCID](https://orcid.org/0000-0001-9151-2607), [e-mail], Delft University of Technology, The Netherlands  <br># DESCRIPTIONThe dataset contains video recordings of human-robot interactions with a Nao robot controlled via the SONAR adaptive control architecture for social norm aware robots. <br>The dataset was collected in the context of the "Human-like norm-aware cognitive robots for autonomous interactions with humans" research project funded by NWO Open Competition Domain Science - XS 21-3 grant. The dataset was generated via experiments with human participants. The experiments were conducted to evaluate the developed adaptive control architecture for social-norm aware robots SONAR. The dataset supplements the article [https://doi.org/10.1007/s12369-024-01172-8]("SONAR: An Adaptive Control Architecture for Social Norm Aware Robots"), where detailed information about the experiments and the related methodologies can be found. <br>In the video recordings, the participants interact with a Nao robot in a casual conversation scenario. Interactions between humans and the robot are entirely autonomous. The participants were instructed to have a conversation with the robot on any topic of their choice for about 10 minutes. During the conversation, they were asked to go through 5 tasks (greeting, role playing game, discussing a personal issue, paying attention to an object, goodbye). Participants could decide on their own when and how to initiate and terminate the tasks. The robot could autonomously interact with the participants thanks to the developed control architecture SONAR which combines several state-of-the-art heories and technologies, including the belief, desire, intention model of reasoning and decision making for rational agents, fuzzy logic theory, and large language models. <br>For every participant, two videos recording are included: - Video of the interaction with the Nao robot controlled via SONAR (Nao-SONAR) - Video of the interaction with the Nao robot controlled via a baseline control architecture (Nao-Chatbot)  <br>Participants were encouraged not to say anything personal during the experiments. Therefore anything that is being said in the video recordings may or not may be true and should be assumed to be a product of fantasy. <br>It is also important to mention that these experiments were carried out in compliance with the General Data Protection Regulation (EU GDPR) 2016/679, and the Delft University of Technology guidance for working with personal data from human participants. Meaning the project was approved by the Human Research Ethics Committee of the Delft University of Technology. In that sense, all participants were informed since the beginning -via an informed consent form- that the videos of their interactions with the social robot would be made available upon request upon the completion of the project.   ## KeywordsHuman-robot interaction, HRI (Human-robot interaction), Human engineering, Videos, Video recordings, Sonar, Robots, Autonomous robots, Social robots, Norm aware robots, BDI (Belief-Desire-Intention), Fuzzy logic, Large language model, LLM (large language model) <br>## Date of data collection2022-12-01 to 2022-12-31  <br>## Date of dataset publication2023-11-30    ## FundingThis research has been supported by the NWO - Open Competition Domain Science XS project "Human-like norm-aware cognitive robots for autonomous interactions with humans" (OCENW.XS21.3.106), which has been financed by the Netherlands Organisation for Scientific Research (NWO).  # FILE OVERVIEWThe dataset contains two video files per participant. The following file naming convention has been used, where &lt;prid\&gt; is a random identifier assigned to a participant:- Video of the interaction with the Nao robot controlled via SONAR (Nao-SONAR): &lt;prid\&gt;_sonar.mp4- Video of the interaction with the Nao robot controlled via a baseline control architecture (Nao-Chatbot): &lt;prid\&gt;_baseline.mp4  <br>Last updated: 2024-02-15

# 基本信息
## 数据集标题
搭载社交规范感知机器人自适应控制架构（SONAR）的NAO机器人人机交互视频记录
### 数据集DOI
https://doi.org/10.4121/50c7a19c-fc0e-4ef3-b35a-dd23bf08470d
## 作者
- **戴维德·德尔安娜博士（Dr. Davide Dell'Anna）**（ORCID标识：https://orcid.org/0000-0002-1162-8341，邮箱：[e-mail]，乌得勒支大学，荷兰，个人主页：https://www.davidedellanna.com）
- **阿妮塔·贾姆希德纳贾德博士（Dr. Anahita Jamshidnejad）**（ORCID标识：https://orcid.org/0000-0001-9151-2607，邮箱：[e-mail]，代尔夫特理工大学，荷兰）

# 数据集说明
本数据集包含搭载SONAR自适应控制架构的社交规范感知机器人NAO的人机交互视频记录。
本数据集源于“面向人类自主交互的类人规范感知认知机器人”研究项目，该项目由荷兰科学研究组织（NWO）开放竞争领域科学-XS 21-3基金资助。数据集通过人类受试者实验生成，实验旨在评估所开发的社交规范感知机器人自适应控制架构SONAR。本数据集配套发表于论文《SONAR: An Adaptive Control Architecture for Social Norm Aware Robots》（DOI: https://doi.org/10.1007/s12369-024-01172-8），实验细节与相关方法论可参阅该论文。

在视频记录中，受试者与NAO机器人在轻松对话场景中展开交互，全程人机交互均为自主运行。受试者被要求围绕任意自选话题与机器人进行约10分钟的对话，期间需完成5项任务：问候、角色扮演游戏、讨论个人话题、关注特定物品、告别。受试者可自主决定任务的发起时机与执行方式。得益于所开发的SONAR控制架构，机器人可与受试者自主交互，该架构融合了多项前沿理论与技术，包括面向理性智能体的信念-愿望-意图（BDI，Belief-Desire-Intention）推理决策模型、模糊逻辑理论以及大语言模型（LLM，Large Language Model）。

每位受试者对应两段视频记录：
- 搭载SONAR控制架构的NAO机器人交互视频（记为Nao-SONAR）
- 基于基线控制架构的NAO机器人交互视频（记为Nao-Chatbot）

受试者被要求在实验中避免提及个人隐私信息，因此视频记录中的对话内容可能为虚构，不代表真实情况。
需特别说明的是，本实验严格遵循《通用数据保护条例（欧盟GDPR）2016/679》以及代尔夫特理工大学关于人类受试者个人数据处理的指导规范。本项目已通过代尔夫特理工大学人类研究伦理委员会的审批。所有受试者在实验初期均通过知情同意书被告知：其与社交机器人的交互视频将在项目完成后按需公开。

## 关键词
人机交互（Human-robot interaction, HRI）、人类工程学、视频、视频记录、SONAR、机器人、自主机器人、社交机器人、规范感知机器人、信念-愿望-意图（BDI，Belief-Desire-Intention）模型、模糊逻辑、大语言模型（LLM，Large Language Model）

## 数据采集时间
2022年12月1日至2022年12月31日

## 数据集发布时间
2023年11月30日

## 资助信息
本研究得到荷兰科学研究组织（NWO）开放竞争领域科学XS项目“面向人类自主交互的类人规范感知认知机器人”（OCENW.XS21.3.106）资助。

# 文件概览
本数据集每位受试者对应两段视频文件，采用如下命名规则，其中&lt;prid&gt;为分配给受试者的随机标识符：
- 搭载SONAR控制架构的NAO机器人交互视频（Nao-SONAR）：&lt;prid&gt;_sonar.mp4
- 基于基线控制架构的NAO机器人交互视频（Nao-Chatbot）：&lt;prid&gt;_baseline.mp4

最后更新时间：2024年2月15日