CAREER: End-User Programming of General-Purpose Robots

机器人可以帮助人类完成日常任务，有可能提高人们的生活质量，使残疾人获得独立。实现这类机器人的一个关键挑战是对它们进行编程，以满足用户独特和不断变化的需求，并在其独特的环境中稳健地发挥作用。大多数机器人学研究都针对这一挑战，试图开发通用或自适应的机器人能力。这种方法取得的成功有限，因为要预测通用机器人的所有可能场景和用例是极其困难的。相反，这项研究的目标是开发可以在部署后在情景中和由最终用户编程的机器人。PI的方法是将方法从广泛的最终用户编程研究领域(以前仅限于个人计算机、智能手机和Web应用程序)应用和扩展到机器人编程。项目成果将产生长期的社会和经济影响，特别是对患有不同程度运动障碍的个人。为了进一步扩大这项研究的影响，该项目将涉及几项外联和教育活动，重点是扩大妇女和残疾个人在计算机科学教育、研究和职业方面的参与。这些活动将直接将项目中开发的工具纳入三个面向K-12学生的教育推广单元和一个本科课程。外展单元将针对三个不同时间尺度的地点；由威斯康星大学DO-IT中心为残疾学生组织的为期一周的夏令营，为期三个小时的“编程女孩”实地旅行练习，以及“编码一小时”活动的一个小时练习。由PI开发的Robotics Capstone课程将进行更新，以整合本项目中开发的工具，并将重点放在有实际客户参与的辅助机器人领域的项目上。所有工具都将在最先进的移动机械手机器人PR2上实现，并将作为开源项目发布。这项研究的一个真正独特的方面是，开发的工具将部署在两个私人拥有的家庭PR2机器人上，导致从真实世界评估中获得第一个洞察力。为了实现她的目标，PI将解决与机器人的不同特征相关的挑战，包括情景、影响环境的能力、移动性、分布式并发处理和整体复杂性。这项研究将在三种编程范例中产生一套新的方法和工具：(I)涉及通过与机器人及其环境直接交互的编程的“情境编程”，(Ii)涉及实际编程但使用高度简化的语言的“简化编程”，以及(Iii)涉及操纵程序实体的抽象的“抽象编程”，针对这些抽象实体的程序被自动综合。所有方法都将被实例化为在多个层次上对机器人进行编程的工具，包括对新知识、技能、任务和规则进行编程。这些项目的主要目标是使这些工具(I)具有足够的表现力来捕捉有用的现实世界任务，(Ii)可供没有技术背景的人使用，以及(Iii)可供不同能力的人使用。第一个目标将通过以用户为中心的家用辅助机器人综合基准的系统实验进行评估，包括对取东西、组织或准备等任务的编程，以及打开或关闭不同家庭中的橱柜、抽屉、电器和水龙头等技能。第二和第三个目标将通过对身体健全和运动障碍的个人进行用户研究来评估。

Robots that can assist humans in everyday tasks have the potential to improve people's quality of life and bring independence to persons with disabilities. A key challenge in realizing such robots is programming them to meet the unique and changing needs of users and to robustly function in their unique environments. Most research in robotics targets this challenge by attempting to develop universal or adaptive robotic capabilities. This approach has had limited success because it is extremely difficult to anticipate all possible scenarios and use-cases for general-purpose robots. Instead, this research aims to develop robots that can be programmed in-context and by end-users after they are deployed. The PI's approach is to apply and extend methods from the broad research area of end-user programming, which has previously been limited to personal computers, smart phones, and Web applications, to programming robots. Project outcomes will have long-term societal and economic impacts, particularly on individuals with varying levels of motor impairments. To further broaden the impacts of this research, the project will involve several outreach and educational activities focused on broadening participation of women and individuals with disabilities in computer science education, research, and careers. These activities will directly integrate tools developed in the project, within three educational outreach modules for K-12 students and one undergraduate course. Outreach modules will target three venues with different time scales; a week-long summer camp for students with disabilities organized by the UW DO-IT center, a three-hour exercise for a "Girls who Code" field trip, and one-hour exercise for an "Hour of Code" event. The Robotics Capstone course developed by the PI will be updated to integrate tools developed in this project and to focus on projects within assistive robotics with the involvement of an actual customer. All tools will be implemented on a state-of-the-art mobile manipulator robot, PR2, and will be released as open-source projects. A truly unique aspect of this research is that developed tools will be deployed on two privately-owned in-home PR2 robots, leading to first insights from real-world evaluations.To achieve her goal the PI will address challenges related to distinct characteristics of robots including situatedness, ability to influence the environment, mobility, distributed concurrent processing, and overall complexity. This research will produce a suite of new methods and tools within three programming paradigms: (i) "situated programming" which involves programming through direct interactions with the robot and its environment, (ii) "simplified programming" which involves actual programming but with highly simplified languages, and (iii) "abstracted programming" which involves manipulating abstractions of program entities for which programs are synthesized automatically. All methods will be instantiated as tools for programming robots at multiple levels, including programming of new knowledge, skills, tasks, and rules. The main objectives of the projects are to make these tools (i) expressive enough to capture useful real-world tasks, (ii) usable by people with no technical background, and (iii) accessible to people with diverse abilities. The first objective will be evaluated through systematic experiments with a comprehensive user-centered benchmark for domestic assistive robots, including programming of tasks such as fetching, organizing, or preparing, and skills such as opening or closing cabinets, drawers, appliances, and faucets in different homes. The second and third objectives will be evaluated through user studies with both able-bodied and motor-impaired individuals.