Designing a socially assistive robot for personalized number concepts learning in preschool children

Designing technological systems for personalized education is an iterative and interdisciplinary process that demands a deep understanding of the application domain, the limitations of current methods and technologies, and the computational methods and complexities behind user modeling and adaptation. We present our design process and the Socially Assistive Robot (SAR) tutoring system to support the efforts of educators in teaching number concepts to preschool children. We focus on the computational considerations of designing a SAR system for young children that may later be personalized along multiple dimensions. We conducted an initial data collection to validate that the system is at the proper challenge level for our target population, and discovered promising patterns in participants' learning styles, nonverbal behavior, and performance. We discuss our plans to leverage the data collected to learn and validate a computational, multidimensional model of number concepts learning.