MRI: Acquisition of Robotic Hardware for Humanoid Research in Cognitive Science and Engineering

该奖项在加州大学默塞德分校建立了一个类人机器人设施，供认知科学家和工程师使用，以调查(1)体现如何限制人类认知模型，(2)人们如何自然地与类人机器人交互，以及(3)机器人控制系统的设计如何最好地解决围绕人-机器人交互的认知问题。该设施的中心将是一个完全移动的类人机器人和一个类人躯干，它配备了两个类似人类的灵巧手臂和一个安装在完全驱动的头部上的视觉系统。尽管关于类人机器人的许多工作都集中在解决与现实世界环境中稳健操作相关的基本工程问题上，但拟议中的设施将支持将机器人用作测试人类认知假说的工具的研究，或者以对人与机器人协调的认知限制敏感的方式增强机器人能力的研究。作为一种科学仪器，类人机器人可以用来向人类实验参与者呈现精确的、可控的运动和交互模式，为探测人类在交互环境中的反应提供了新的方法。这种机器人的感知和运动能力也使其成为人类认知过程计算模型的一个具有挑战性的试验台。开发适当支持人-机器人交互的认知系统将具体化和检验我们对具身感知、类人运动协调和合作互动的理解。拟议的设施还旨在支持认知科学和工程学之间的跨学科研究。新兴的关于人-机器人交互的认知研究有可能改变机器人学的研究，并导致仿人机器人控制系统设计的实用创新。这些额外的工程贡献将是极其宝贵的，因为机器人将配备适合人类环境的物理和认知能力，以及适合人与机器人协作的沟通和协调技能。事实上，机器人助手已经在许多应用领域显示出潜力，包括物理治疗、照顾老年人和残疾人、协作工作以及太空探索期间的宇航员支持。尽管最近的技术创新使模仿人体的可靠和负担得起的机器人机制的开发成为可能，但与这些机构的智能控制相关的许多问题仍然存在。通过支持在机器人框架内探索人类认知模型，并将认知科学的见解普遍应用于体感和运动控制问题，预计将在杂乱环境中的运动、空间意识和空间推理、灵活的物体操作、任务导向的关注和感知、模仿和学习、终身适应以及多模式社会互动和与人类的协调(包括使用自然手势)等领域取得实际进展。该设施将推动加州大学默塞德分校的综合研究和教学项目的发展，该校区位于教育水平较低的经济困难地区。这种设备的提供将使学生获得在认知科学、人工智能、运动规划、计算机图形学、计算机动画、计算机视觉、机器学习、机器人算法和类人机器人方面的独特培训机会。

This award establishes a humanoid robotics facility at the University of California, Merced to be used by cognitive scientists and engineers to investigate (1) how embodiment constrains models of human cognition, (2) how people naturally interact with humanoid robots, and (3) how the design of robotic control systems can best address the cognitive issues surrounding human-robot interaction. The centerpieces of this facility will be a fully mobile humanoid robot and a humanoid torso equipped with two human-like dexterous arms and a vision system mounted on a fully actuated head. Though much work on humanoid robots has focused on solving fundamental engineering problems associated with robust operation in real world environments, the proposed facility will support research that uses the robot as an instrument to test hypotheses of human cognition or that augments robotic capabilities in a manner sensitive to the cognitive limitations of human-robot coordination. As a scientific instrument, a humanoid robot can be used to present precise, controlled motions and patterns of interaction to human experimental participants, offering new methods for probing human responding in interactive contexts. The embodied perceptual and motor capabilities of such a robot also make it a challenging testbed for computational models of human cognitive processes. Developing cognitive systems that appropriately support human-robot interaction will reify and test our understanding of embodied perception, humanoid motor coordination, and cooperative interaction.The proposed facility is also intended to support interdisciplinary research at the boundary between cognitive science and engineering. Emerging cognitive research on human-robot interaction has the potential to transform robotics research and result in practical innovations in the design of humanoid robot control systems. These additional engineering contributions will be extremely valuable, as robots will be equipped with physical and cognitive abilities that are appropriate for human environments, and with communication and coordination skills that are appropriate for human-robot collaboration. Indeed, robotic assistants have already demonstrated potential in numerous application areas, including physical therapy, care for the elderly and disabled, collaborative work, and astronaut support during space exploration. Though recent technological innovations have allowed for the development of reliable and affordable robotic mechanisms that mimic human bodies, many questions related to the intelligent control of these bodies remain. By supporting the exploration of models of human cognition within a robotic framework, and by generally applying insights from cognitive science to the problems of embodied perception and motor control, practical progress is expected in domains such as locomotion in cluttered environments, spatial awareness and spatial reasoning, dexterous object manipulation, task-oriented attention and perception, imitation and learning, life long adaptation, and multi-modal social interaction and coordination with humans, including the use of natural gestures.This facility will advance the development of integrated research and teaching programs at UC Merced, a new campus located in an economically challenged region with low educational levels. The availability of this equipment will grant students access to unique training opportunities in Cognitive Science, Artificial Intelligence, Motion Planning, Computer Graphics, Computer Animation, Computer Vision, Machine Learning, Robot Algorithms, and Humanoid Robotics.