Developing a Semi-autonomous Robot to Engage Children with Special Needs and Their Peers in Robot-Assisted Play

摘要

Despite the wide variety of robots used in human-robot interaction (HRI) scenarios, the potential of robots as connectors whilst acting as play mediators has not been fully explored.
\n
\nRobots present an opportunity to redefine traditional game scenarios by being physical embodiments of agents/game elements. Robot assisted play has been used to reduce the barriers that children with physical special needs experience. However, many projects focus on child-robot interaction rather than child-child interaction. In an attempt to address this gap, a semi-autonomous mobile robot, MyJay, was created. This thesis discusses the successful development of MyJay and its potential contribution in future HRI studies.
\n
\nMyJay is an open-source robot that plays a basketball-like game. It features light and color for communicative feedback, omni-directional mobility, robust mechanisms, adjustable levels of autonomy for dynamic interaction, and a child-friendly aesthetically-pleasing outer shell.
\n
\nThe design process included target users such as children with special needs and therapists in order to create a robot that ensures repeated use, engagement, and long-term interaction. A hybrid approach was taken to involve stakeholders, combining user-centered design and co-design, exemplifying that children can be included in the creation process even when it is not possible to hold in-person co-design sessions due to COVID-19.
\n
\nAside from the care taken to meet user requirements, the robot was designed with researchers in mind, featuring extensible software and ROS compatibility. The frame is constructed from aluminum to ensure rigidity, and most functional parts related to gameplay are 3D printed to allow for quick swapping, should a need to change game mechanics arise. The modularity in software and in mechanical aspects should increase the potential of MyJay as a valuable research tool for future HRI studies. 
\n
\n
\nFinally, a novel framework to simulate teleoperation difficulties for individuals with upper-limb mobility challenges is proposed, along with a dynamic assistance algorithm to aid in the teleoperation process.