A humanoid robot assistant for the classification of students according to their type of dysgraphia

Students diagnosed with dysgraphia, a learning disability that affects handwriting and fine motor skills, are unable to follow the learning process correctly. The majority of regular teachers cannot support the correct learning of these students. Our present work aims to develop and implement a knowledge-learning model for a robot assistant to provide a diagnosis to dysgraphic students, in order to categorize them according to their types of dysgraphia. The conception of the general process of categorization is based on exploiting the expertise of specialists in the field in order to design a matrix of correspondence between symptoms and types of dysgraphia. This correspondence provides an automated technique for the efficient formation of decision trees. A decision tree classification is a popular approach to the problem, resulting in an accurate, flexible, and efficient classification. Our classification strategy has reached an advanced level; it uses a severity filtering procedure to divide dysgraphic learners into 3 categories: dysgraphic learners with mild dysgraphia, dysgraphic learners with a moderate type of dysgraphia, and dysgraphic learners with a severe type of dysgraphia. Therefore, the results indicate that 92.3 % of the finding of our assistant robot's diagnosis are similar to the finding of the experts' diagnosis. The results indicate that the robot was able to classify dysgraphic students according to their type of dysgraphia, giving the degree of severity that a learner attains in that type so that the human tutor can assign the students to the appropriate rehabilitation program depending on their specific needs.