Extraction of Peer-pressure Behaviors that Change Low-moral Behaviors from Large Amount of Data and Its Application to Robots

从大量数据中提取改变低道德行为的同伴压力行为及其在机器人中的应用

• 批准号: 22KJ1751
• 负责人: DU KANGHUI
• 金额: $ 1.41万
• 依托单位: Kyoto University
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for JSPS Fellows
• 财政年份: 2023
• 资助国家: 日本
• 起止时间: 2023-03-08 至 2024-03-31
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-22KJ1751/
• 关键词: Human-robot Interaction; Low-moral Behaviors; Low-Morarl Behaviors; Real-World Dataset; Human-Human Interaction; Human-Robot Interaction

In the last year we have updated our data collection system and collected more data in the public space. By observing and analyzing the data from the real-world we have better understood the low-moral behaviors. In order to extract stable 3D information from humans, we updated the system by developing a new data extractor.We have done an analysis from the data we have collected in the public space and found a common reaction that humans do to low-moral behaviors. We have extracted tens of examples from human-human interaction cases and proposed a new robot model called “suggestive avoidance”. We further conducted a lab study to test how the new model could influence people. And the model successfully made people perceive the robot as more being bothered in the result.

去年，我们更新了我们的数据收集系统，并在公共空间收集了更多数据。通过观察和分析来自现实世界的数据，我们更好地理解了道德失范行为。为了从人类身上提取稳定的3D信息，我们通过开发新的数据抽取器对系统进行了更新。我们对在公共空间收集的数据进行了分析，发现人类对道德败坏行为的普遍反应。我们从人与人交互的案例中提取了数十个例子，并提出了一种新的机器人模型--暗示回避。我们进一步进行了一项实验室研究，以测试新模式如何影响人们。该模型成功地让人们感觉到机器人在结果中更受困扰。