NRI: FND: End-User Robot Programming

点击翻译按钮获取中文摘要

Collaborative robots, or robots that can safely work in close proximity to humans, promise to revolutionize society. Such robots could help the country remain competitive in the service and manufacturing industries. But this vision has not yet been achieved because programming robots remains difficult, time consuming, and, most importantly, only feasible by experts after years of training. This project seeks to remove this final barrier to progress by making collaborative robots programmable by their owners. That is, without any special training, to empower small business owners, lab technicians, and small-scale manufacturers to be able to program a robot to do meaningful tasks. Project outcomes will not only benefit American businesses by lowering costs and freeing human workers to complete higher-level tasks, they will also improve access to technology for many. Because the research will be conducted at a highly diverse university, in collaboration with a rural community college and a rural manufacturing research center, additional broad impacts will derive from exposing a diverse group of students to cutting edge technology they would otherwise never know existed. Furthermore, by conducting experiments and demonstrations at both rural sites the team will impart to the broader community a vision for future career choices while driving robot programming technology to the next level.Currently, the high cost of programming collaborative robots remains a major barrier to adoption. This project will create a suite of end-user programming languages that make common robot programming tasks accessible to novices, dramatically reducing this barrier to adoption. The work is based on two key insights. First, because collaborative robots offer unprecedented safety features, end-users can safely operate and program them. Second, by leveraging the last twenty years of progress in educational programming languages (e.g., MIT’s Scratch), interfaces can be created that support the programming of realistic tasks yet can be learned in minutes. The project will comprise four thrusts: (1) creation of a block-based robot programming environment that enables a large set of realistic tasks; (2) development of an in-line tutorial system for learning block-based programming without formal training; (3) implementation of a parallel programming language for multi-armed programming; and (4) definition of a multi-threaded programming approach for maximizing robot usage. This comprehensive set of tools will enable end-users to tackle many common robot programming tasks that are currently out of their reach. To ensure that the various approaches actually help end-users, they will be evaluated through extensive user studies run both on an urban and rural university campus. These studies, where subjects complete realistic tasks with the prototypes, will have three outcomes. First, preparing for the studies naturally hardens the research prototypes into high-quality software, allowing accurate evaluation. Second, the study will generate quantifiable results on the effectiveness of end-users programming with each tool. Third, through video analysis and interviews additional insight will be derived into further issues that hinder end-users, thereby laying the groundwork for the next round of innovations.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.