Smart Science Exhibits: Enhancing STEM Learning through Sustained Use of Mixed-Reality in Multiple Informal Learning Spaces

博物馆和类似的非正式学习场所为儿童和家庭提供了以一种引人入胜的方式一起学习的机会。目前的展品主要依靠家长、老师、标牌和科技馆的工作人员提供支持和指导。由于拥有知识渊博的工作人员并不总是可行的，而且并不是所有的父母都有相同的知识和背景，孩子们得到的支持多种多样，人们往往错过了学习经历或活动的重点。该项目将开发和研究一种新的智能科学展品类型，在自适应系统中使用人工智能(AI)来支持儿童通过做科学来学习科学。该项目的目的是结合人工智能的适应性和个性化，以最大限度地提高基于探究的STEM学习和非正式学习环境中的参与度。这项研究建立在项目团队的第一个智能科学展览(EarthShake)的基础上，该展览使用人工智能视觉根据参观者的行动向他们提供互动反馈，并引导他们进行科学探究。在该项目的初步工作中，第一个智能展览展示了比传统博物馆展览更高的参与度和更多的学习收获，这些展览涉及相同的科学内容。智能展品可以扩展和增强工作人员和家长可以提供的有限支持。该项目将开发和研究混合探索和人工智能指导的自适应方法，这将在建设性的探索过程中使反馈个性化。该项目将建立在学习科学技术和技术的基础上，在正式环境下的智能辅导系统中得到验证，并将其应用于不同的非正式学习环境。其目标是提供即时学习支持，这将延长参观者在展品上花费的时间，从而深化基于探究的科学学习。该项目正在与科学博物馆和课外项目合作，这将使成千上万来自不同背景的儿童和家庭每年都能使用该项目的智能展品。Smart Science Exhiits由推进非正式STEM学习(AISL)计划资助，该计划支持创新研究、方法和资源，作为其加强非正式环境中学习的整体战略的一部分。许多非正式学习环境正在考虑混合现实(MR)技术，以增加对科学的参与和理解。利用Smart Science展品，该项目将调查混合现实系统中的设计选择如何影响用户对STEM概念的参与和学习。(混合现实是物理世界和数字世界的融合，实现了人类和人工智能之间的互动。)项目研究将通过调查关于学习者结果的经验结果来扩展目前的研究，这在很大程度上是描述性的。关键的研究问题是：哪些类型的适应性和个性化可以总体上改进Smart Science展品和MR系统？什么是探索和AI指导的平衡，才能最大限度地享受、参与和学习？关于智能科学展览有效特征的研究结果是否适用于不同的内容领域和非正式学习环境？该项目将采用以用户为中心的设计研究、形成性评估和受控实验，以发现混合现实系统应该如何设计，以最好地满足非正式学习环境中的游客和工作人员的需求，包括多个博物馆和课后提供者。在各种非正式环境中，学习者在混合现实体验中的行为数据将为Smart Science展品的设计提供信息。该项目将调查适应性方法是否适用于所有内容和背景，以实现更好的STEM学习、参与、协作和富有成效的对话。该项目将纳入该团队之前的技术研究，该研究开发了跟踪儿童身体互动的视觉技术和互动教学技术，为儿童的询问和构建行为提供支架和反应反馈。新的技术研究将开发人工智能技术，用于自适应任务选择和个性化反馈，利用访问者的互动历史。项目研究和设计资源将通过在年度会议上的陈述与科学博物馆教育工作者和设计师广泛共享，并通过同行评议的期刊出版物和专业出版物与研究人员、开发人员和其他人共享。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Museums and similar informal learning settings offer opportunities for children and families to learn together in an engaging way. Current exhibits rely mainly on parents, teachers, signage, and staff in science museums to provide support and guidance. Since it is not always feasible to have knowledgeable staff on hand and not all parents have the same knowledge and background, children receive varied support and people often miss the point of the learning experience or activity. This project will develop and research a new genre of Smart Science Exhibits that use artificial intelligence (AI) in an adaptive system to support children in learning science by doing science. The aim of the project is to incorporate AI adaptivity and personalization to maximize inquiry-based STEM learning and engagement in informal learning settings. This research builds on the project team’s first Smart Science Exhibit (EarthShake), which uses AI vision to give interactive feedback to visitors based on their actions and guides them through scientific inquiry. In the project’s preliminary work, the first smart exhibit demonstrated higher engagement and more learning gains than resulted from a traditional museum exhibit addressing the same scientific content. Smart exhibits can extend and enhance the limited support that staff and parents can provide. This project will develop and investigate adaptive approaches to mixing exploration and AI guidance, which will personalize feedback during constructive exploration. The project will build on learning science techniques and technology, proven in intelligent tutoring systems in formal settings, and apply this to different informal learning contexts. The goal is to provide just-in-time learning support, which will extend the time visitors spend with exhibits, thereby deepening inquiry-based science learning. The project is partnering with science museums and afterschool programs, which will enable thousands of children and families from a wide variety of backgrounds to use the project’s smart exhibits each year. Smart Science Exhibits is funded by the Advancing Informal STEM Learning (AISL) program which supports innovative research, approaches, and resources as part of its overall strategy to enhance learning in informal environments. Many informal learning settings are considering mixed-reality (MR) technologies to increase engagement and understanding of science. Using Smart Science Exhibits, the project will investigate how design choices in mixed-reality systems impact users’ engagement and learning of STEM concepts. (Mixed reality is the blending of the physical world and the digital world, enabling interaction between human and artificial intelligence.) Project research will extend current research, which is largely descriptive, by investigating empirical results on learner outcomes. Key research questions are: What types of adaptivity and personalization can improve Smart Science Exhibits and MR systems generally? What balance of exploration and AI guidance is best to maximize enjoyment, engagement and learning? Do findings about the effective features of Smart Science Exhibits generalize to different content areas and informal learning settings? The project will employ user-centered design research, formative evaluation, and controlled experimentation to discover how mixed-reality systems should be designed to best meet visitor and staff needs in informal learning settings including multiple museums and afterschool providers. Data on learner behaviors in mixed-reality experiences in a variety of informal settings will inform the design of Smart Science Exhibits. The project will investigate whether adaptive approaches generalize across content and context to achieve better STEM learning, engagement, collaboration, and productive dialogue. The project will incorporate the team’s prior technical research, which developed both vision techniques to track children’s physical interactions and interactive pedagogical techniques to provide scaffolds for and reactive feedback on children’s inquiry and construction behaviors. New technical research will develop AI techniques for adaptive task selection and personalized feedback that draws on a visitor’s history of interaction. Project research and design resources will be widely shared with the science museum educators and designers through presentations at annual conferences and with researchers, developers and others through peer-reviewed journal publications and professional publications.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.