Implementación de un sistema apoyo a la educación parvularia con un robot humanoide utilizando técnicas de visión artificial.

Abstract

In this research work, a support system for preschool education was implemented with a humanoid robot using artificial vision techniques called Happy, the Nao robot was selected which met the operational and functional requirements at hardware and software level. The system consisted of teaching colors and geometric shapes to 4-year-olds through the use of flashcards, for this artificial vision techniques were used such as: preprocessing, segmentation and feature extraction, where the algorithms of supervised learning were trained with convolutional neural networks and classifiers of HARR type in a cascade network core; with this, the system had the ability to identify images by color and geometric shape. Based on tests of color detector, shape detector and colored-shape detector in controlled and uncontrolled environments, it was determined that trained models are 95% accurate in squares, 92% in circles, and 91% in triangles and an acceptance of the Nao robot of 95.12% by the children. It is concluded that the system Happy is helpful in the cognitive training of children as it is a friendly tool that captures their attention, trains their visual memory and arouses their curiosity. It is recommended to keep conducting research that involves robotics in education, to innovate the pedagogical curriculum of educational institutions in the future.