The main purpose of the present WP is to provide the computational algorithms and models necessary to both:

• Characterise the child’s behaviours.
• Assess quantify the children’s degree of motivation/engagement as well as their performance. The specific challenge of the present work package in this respect is therefore to infer these values from the behaviours previously identified.

The outputs from the algorithms thus developed will feed into the social cognitive controller for robot behaviour selection (WP6).

A secondary purpose of this work package is to present the collected data in a manner accessible to therapists for the purpose of assisting diagnosis offline (i.e. after the therapy session itself), comparing the progress over different interactions as well as providing new diagnostic methods in the future.