Contributions to the data adquisition technologies and the methods for automatic detection and classification of activities in a smart toys environment

Abstract

The development of novel Internet services and sensor technologies has allowed the appearance of new opportunities for the development of tools for the acquisition and analisys of data in various fields. The possibility of integrating smaller and with lower energy consompution sensors makes possible to use these sensor-based systems in scenarios where it was formerly not possible. One of these scenarios is the healthcare related systems, where important gains in terms of users’ quality of live can be obtained. As a drawback, these systems raise new challenges related with the security and privacy of the data managed in them. The assessment of children development through games or other similar playing activities is a well-studied field. Through time, experts in this type of evaluation (psychologists, physiotherapists, pedagogues, etc.) have created and improved regulated scales which allow them to detect, under certain circumstances, possible development difficulties in children from early ages. The early detection of these difficulties is crucial for its monitoring and possible treatment to be as effective as possible. Nowadays, the activities used for development assessment require the manual and in-person data compilation in the course of the game, which makes more difficult the analysis and obtaining conclusions from experts. The research carried out during the development of this Thesis has been focused on the study of possibilities for offering technological solutions to improve the playing activities used in the early detection of children development difficulties. To achive this goal, we have addressed both general issues such as the creation of a specific Smart Toys platform, and more specific issues such as the development of methodologies for the autonmous and automatic analysis of each movement composing an activity. The platform has been developed following the Internet of Things (IoT) paradigm, aiming to integrate in it every actor involved in the assessment (from the design of sensor-embedded toys to the definition of the communications between devices, and the appropriate security and privacy mechanisms). The use of these technologies and methods can provide a significative improvement in terms of development assessment metrics measurement accuracy and can also opens the possibility of obtaining new measurement parameters which could not be obtained manually. Three specific objectives have been defined in this Thesis: The first one is the design of the platform based on an IoT architecture, and the Smart Toys composing it. To achieve this objective, we have studied the distinctive features of Smart Toys and we have proposed specific solutions for the challenges raised by their integration in an IoT enviornment. The second objective is the design and development of custom security and privacy mechanisms which offer the neccesary protection during the sensible data exchange in the platform. Finally, the third objective is focused on the creation and development of methodologies for the detection and classification of movements performed with the Smart Toys, in order to obtain valuable information to be used by the children development experts.