iReCHeCk - A Robotic Handwriting Companion – iReCHeCk

Handwriting disorders (termed dysgraphia) are a far from a singular problem as nearly 6% of children between 5 and 11 years in France is considered dysgraphic. Moreover, research highlights the fundamental importance to detect and remediate these handwriting difficulties as soon as possible in children' educational path. At the moment, the detection of dysgraphia is performed through a standard test called "Concise Evaluation Scale for Children’s Handwriting" (BHK). This detection is laborious because of its high cost and subjectivity. In order to establish the diagnosis of dysgraphia, therapists investigate the social context of the child's difficulties (i.e. stress at school, anxiety), and use observable metrics such as handwriting tests (i.e. BHK), analysis of child's cahier and posture. Making the child to take these tests can take up to 3 hours and are aimed to understand and to provide adapted training to the child. After the diagnosis, the therapeutic management consists in a long series of training sessions to help the child to gain better fine motor skills. These sessions can be affectively challenging for children who often lack confidence and motivation to practice their handwriting. Previous works conducted at the CHILI Lab (EPFL) tended to show that the learning by teaching approach using a robotic co-learner agent had positive effects on the child's extrinsic motivation allowing to engage children in the handwriting task for longer sessions. With the project iReCHeCk, we propose to extend the set of observable metrics used by therapists with low level handwriting and body posture features characterizing non-observable micro-movements important in the handwriting process. In a first phase, we will investigates new features that can help the diagnostic and the remediation over two channels, handwriting pen traces and child postural behaviors, extracting characteristic of handwriting skills and attitude within a co-learning activity with a robot. We will exploit cutting-edge Machine Learning techniques that allow to give explainable outputs to the therapist. A second phase will aim to develop engaging training activities in order to 1) monitor the learning status of the child and allow the therapist to constantly evaluate and quantify the progresses, and 2) adapt the robot's attitude and the training task to the needs of the learner. These activities will be tested on two populations of children suffering of dysgraphia : 1) Typically Developed Children that are learning how to write at school, 2) Children with Neuro-Developmental Disorders that are hospitalized or followed on a weekly basis by a therapist and that present handwriting difficulties along with other disorders (i.e. attentional, autistic) This validation on two populations will allow us to frame our findings both pedagogically and clinically. Finally, with this project, by investigating non-observable low level features of handwriting, we aim to provide new insight that can lead to further understanding of the underlying causes of dysgprahia.