Individualized Adaptive Robot-Mediated Intervention Architecture for Autism

项目摘要：本研究提出了一种新型的变革性机器人干预技术，称为ARIA（Adaptive Robot-mediated intervention Architecture，自适应机器人介导干预架构），它具有加速自闭症谱系障碍（ASD）幼儿社会沟通技能发展的潜力。ARIA将流畅地集成一个人形机器人、多个空间分布的摄像头网络、一系列显示监视器，以及复杂但高效的面部、凝视和手势检测方法，以创造一个高度灵活和自适应的智能环境，潜在地提高ASD幼儿的早期联合注意力和模仿相关技能。该系统的应用将在两项用户研究中进行检验，这些研究以明确定义的自闭症儿童为样本，为机器人干预的推广和潜在影响的重要问题提供具体的答案和方向。智力优势：提出的研究推进了智能自适应机器人平台的设计和开发，为患有ASD的幼儿提供了潜在的变革性干预应用。这里提出的具体技术创新有可能为新的非侵入性和闭环人机交互学习范式做出重大贡献，并有可能广泛扩展到具有大量神经发育条件和限制感官脆弱性的个体。从机器人科学和技术的角度来看，该项目将有助于设计和开发智能学习环境，自适应机器人的智能系统架构以及动态人机交互的情感计算和控制。特别是，它有潜力为情感计算，特别是由非侵入性凝视和注意处理介导的情感计算，开发新的有效的计算方法的应用做出重大贡献。它还将通过开发机器人手势识别和自适应响应的新方法，为基于手势的闭环人机交互做出贡献。该项目将开发一个框架和工具来设计自适应环境，以增强机器人和具体化的社会互动，智能地、流畅地将注意力和手势信息的实时行为指数集成到灵活可控的响应系统中。简而言之，所提出的活动代表了一个系统具有从根本上推进智能人机交互的工程知识的潜力。这种范式也可能潜在地影响我们对自闭症谱系障碍干预本身的科学理解。嵌入式用户研究将测试机器人干预对ASD早期核心症状的潜在功效。更广泛的影响：根据疾病控制和预防中心（CDC）的最新患病率估计，自闭症儿童的患病率为1 / 88，有效的早期识别和治疗通常被视为突发公共卫生事件。ASD在整个生命周期中的成本被认为是巨大的，最近的个人增量生命周期成本预测超过320万美元，全国成本每年超过350亿美元。提出的研究明确侧重于实现具有改善早期ASD相关损伤潜力的机器人干预技术，并可能对这一人群产生显著的有益影响。这项研究可能会进一步发展一种技术，该技术可以使有效干预的所有核心组成部分仅以典型干预计划的一小部分成本实现，同时提高干预提供者系统控制和促进针对个体缺陷的干预相关技能的能力。这些教育活动将训练和指导本科生和研究生进行拟议的研究，并通过几门课程将研究带入课堂。拓展活动将包括为高中生提供研究机会，特别是目前在STEM（科学、技术、工程和数学）领域代表性不足的群体，并在夏季为高中教师提供研究经验。该项目通过正式传播给自闭症谱系障碍家庭、临床和科学界，提供了强大的社区联系。

PI: Sarkar, Nilanjan and Warren, ZacharyProposal Number: 1264462Project Summary: A novel and transformative robotic intervention technology, called ARIA(Adaptive Robot-mediated Intervention Architecture), with the potential to accelerate social communication skill development for young children with autism spectrum disorders (ASD) is proposed in this research. ARIA will fluidly integrate a humanoid robot, multiple spatially distributed network of cameras, an array of display monitors, as well as a complex but efficient computational face, gaze and gesture detection methodology in order to create a highly flexible and adaptive intelligent environment to potentially advance early joint attention and imitation related skills for young children with ASD. Application of this system will be examined across two user studies with well-defined samples of young children with ASD to provide specific answers and direction to important questions of generalization and potential impact of robotic intervention.Intellectual Merit: The proposed research advances the design and development of intelligent adaptive robotic platforms to offer a potentially transformative intervention application for young children with ASD. The specific technological innovation proposed here has the potential to significantly contribute to new non-invasive and closed-loop human-robot interaction learning paradigms with potential broad extension to individuals with a vast array of neurodevelopmental conditions and limiting sensory vulnerabilities across the lifespan. From the perspective of the science and technology of robotics, the project will contribute towards the design and development of smart environments for learning, intelligent system architecture for adaptive robotics as well as affective computing and control of dynamic human-robot interaction. In particular, it has the potential to significantly contribute towards developing novel efficient applications of computational methods for affective computing, particularly affective computing mediated by non-invasive gaze and attention processing. It will also contribute towards closed loop gesture-based human-robot interaction by developing new methodologies for gesture recognition and adaptive response from the robot. The project will develop a framework and tools to design adaptive environments for enhanced robotic and embodied social interaction that intelligently and fluidly integrates real-time behavioral indices of attentive and gesture information into flexible and controllable response systems. In short, the proposed activity represents a system has the potential to fundamentally advance the engineering knowledge of intelligent human-robotic interaction. This paradigm may also potently impact our understanding of the science of ASD intervention itself. The embedded user studies will test the potential efficacy of robotic intervention on the earliest core symptoms of ASD.Broader Impacts: With the most recent Centers for Disease Control and Prevention (CDC) prevalence estimates for children with ASD at 1 in 88, effective early identification and treatment is often characterized as a public health emergency. The costs of ASD are thought to be enormous across the lifespan, with recent individual incremental lifetime cost projections exceeding $3.2 million and national cost over $35 billion annually. The proposed research explicitly focuses on realizing robotic intervention technologies with potential for improving early ASD related impairments and could have significant beneficial impact on this population. This research may further a technology that can enable all core components of effective intervention at only a fraction of the cost of typical intervention programs, while at the same time increasing the ability of the intervention provider to systematically control and promote intervention related skills targeting individual deficit. The educational activities will train and mentor undergraduate and graduate students in the proposed research, and bring research into classroom through several courses. The outreach activities will include offering research opportunities to high school students, especially among groups currently underrepresented in STEM (science, technology, engineering, and mathematics) fields, and providing high school teachers with research experience during summer. The project offers a strong community connection through formal dissemination to ASD family, clinical, and scientific communities.