Biologically Inspired Robotics: Intelligent Systems for Trustworthy Human-Robot Co-learning and Adaptation

仿生机器人：可信赖的人机协同学习和适应的智能系统

• 批准号: RGPIN-2019-05223
• 负责人: Dautenhahn, Kerstin
• 金额: $ 3.5万
• 依托单位: University of Waterloo
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2020
• 资助国家: 加拿大
• 起止时间: 2020-01-01 至 2021-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=715423
• 关键词: Biologically; Inspired; Robotics; Intelligent; Systems

Human-robot interactions are increasingly important in applications where intelligent systems provide assistance to or work alongside people. The ultimate goal of this research program is to develop novel biologically-inspired mechanisms for intelligent system design, enabling intelligent, autonomous robots to engage with people in joint-tasks facilitating co-learning and co-adaptation. Specific objectives over the next 5 years of this program are a) the development of appropriate co-learning computational mechanisms for trustworthy robots that learn and adapt in real-time, b) the study of the impact of robot embodiment, errors, and other factors relevant in establishing, maintaining and repairing human-robot trust, and integrating those factors into computational architectures that can sustain long-term human-robot interaction, c) design of computational mechanisms for robots that can assess and convey (e.g. visualise) their level of competence and confidence in their own learning and decision making, in order to improve human-robot trust, and d) to develop computational, behavioural models for creating trustworthy robot behaviour in terms of the robot's use of social cues and interaction modalities. Experimental evaluation of the architectures and algorithms will involve technical evaluation as well as validation in user studies. Outcomes of this work are expected to change the way social and intelligent autonomous robots interact with people. Methodologically, my work is using diverse methods from Artificial Intelligence, Human-Computer and Human-Robot Interaction. A novelty of this research program is to take inspiration from human/animal behaviour and then design computational models for social and intelligent robots. Biological models of communication and interaction are more universally and readily understood, which can yield more robust and efficient means of communication, and to generalize better across different robot embodiments, compared to the common focus on humanoid/ android robots that evoke high expectations in people in terms of the system's capabilities, but often lead to disappointment when expectations are not met. Outcomes of this program will significantly advance computer science and engineering towards creating robotic systems that people can “naturally” and efficiently interact with, and that they will trust. The project will train HQP in developing fundamental solutions to the development of trustworthy, social and intelligent robots. It will significantly advance the students' interdisciplinary knowledge and skills needed for a future career in academia or industry, where project results can potentially transform the use of such systems in real-world application areas, e.g. the care or service sectors in Canada.

人机交互在智能系统为人类提供帮助或与人类一起工作的应用中越来越重要。这项研究计划的最终目标是为智能系统设计开发新的生物启发机制，使智能、自主的机器人能够与人一起参与联合任务，促进共同学习和共同适应。该计划未来5年的具体目标是：a)为实时学习和适应的可信赖机器人开发适当的共同学习计算机制；b)研究机器人体现、错误和其他与建立、维护和修复人机信任相关的因素的影响，并将这些因素整合到能够维持长期人机交互的计算架构中。C)设计机器人的计算机制，可以评估和传达（例如可视化）他们在自己的学习和决策中的能力水平和信心，以提高人与机器人的信任，d)开发计算行为模型，以创建可信赖的机器人行为，就机器人使用社会线索和交互方式而言。架构和算法的实验评估将包括技术评估以及用户研究中的验证。这项工作的成果有望改变社交和智能自主机器人与人互动的方式。在方法论上，我的工作使用了人工智能、人机交互和人机交互等多种方法。这个研究项目的新颖之处在于从人类/动物行为中获得灵感，然后为社交和智能机器人设计计算模型。通信和交互的生物模型更普遍，更容易理解，这可以产生更强大和有效的通信手段，并在不同的机器人实施例中更好地推广，相比之下，人形/机器人的共同焦点唤起了人们对系统能力的高期望，但往往导致失望当期望没有得到满足。这个项目的成果将显著推动计算机科学和工程朝着创造机器人系统的方向发展，使人们能够“自然”、有效地与之互动，并得到人们的信任。该项目将培训HQP开发可信赖、社交和智能机器人的基本解决方案。它将大大提高学生在学术界或工业界未来职业所需的跨学科知识和技能，项目结果可能会改变这些系统在现实世界应用领域的使用，例如加拿大的护理或服务部门。