I-Corps: A Do-it-yourself (DIY) Social Robot Platform for Science, Technology, Engineering, Arts and Mathematics (STEAM) Education

点击翻译按钮获取中文摘要

The broader/commercial impacts of this I-Corps project is the development of a robotics platform that teaches the technical and creative aspects of engineering through play-based learning. The multifaceted construction process that combines robotics with crafts conveys that engineering draws from humanistic fields, empowering users from diverse backgrounds to get involved in robot-building. Beyond the robot’s physical construction, the interactive, behavior-authoring process allows users to create robot behaviors without requiring programming experience, while also learning about the data-driven mechanisms underlying the robot’s behavior model. This intuitive play-based interaction teaches the influence of user-provided data on machine learning systems that are becoming increasingly ubiquitous in society (e.g., recommendation systems and advertisements). The project empowering users to become more cognizant of the role that robotics systems play in their lives. As a whole, the accessibility of the platform, from its do-it-yourself design to the intuitive programming of its behaviors, allows non-expert users to learn the various skills used in robot-building.This I-Corps project is based on the development of an accessible robotic construction platform. The process of designing and creating robots is intensive and time-consuming. The construction projects are often tailored towards robotics researchers and specific use cases. The proposed technology is an accessible and replicable robot platform that is easy to assemble and able to be customized for different applications. The technology’s mechanical structure is made from flat components that are easy to manufacture and assemble and uses minimal hardware throughout its construction. Creating robot behaviors has traditionally relied on trained roboticists to manually design preprogrammed actions. To address the repetitiveness of these canned responses and the difficulty for end users to author behaviors, a behavior generation workflow was developed that enables users to develop new robot behaviors without a programming background. Users control the robot’s movements through a smartphone application, which allows them to build a dataset of behaviors to train a neural network model. This model may generate new realistic behaviors, modify emotive features of existing behaviors, and/or translate user inputs such as facial expressions and gestures into robot movements. This technology furthers the accessibility of the platform by enabling the users themselves to design the robot’s appearance and behaviors.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.