The Structuring of the Self Through Relational Patterns of Movement Using Data from the Microsoft Kinect 2 to Study Baby-Caregiver Interaction

Abstract

In this paper we will illustrate the progress of a research that intends to use computational methods to analyze the complex interaction between a child and his caregiver, focusing on motor aspects. According to the theory of Ruella Frank, Gestalt psychotherapist, the child acquires motor patterns through the interaction with the caregiver that will be repeated over the course of life. Six fundamental movements have been identified and defined, which allow the child to develop self-awareness. The objective of the study is the creation of an observation tool of dyadic interaction through the automatic recognition of these movements whose purpose is the identification and differentiation of the peculiar vocabulary of movements that is established between parent and child. Thanks to the use of Kinect v2 and through the Kinect Studio program, we have trained a network to identify three of the six basic movements: “Reaching”, “Pushing” and “Pulling”. The results of this training show that the machine discriminates the gestures that have been defined and recognized through different features, although they visually share a part of the progression. The future perspectives of this study foresee to apply the methodology described in this paper to the other three gestures, widen the database and refine the tagging process, to obtain more precise results. The ultimate aim is to include the database in an application that we have proposed to develop, useful for the recognition of the peculiar dyadic baby-caregiver movements.