Di Mauro, Dario (2017) Human-Computer Interaction in Intelligent Environments. [Tesi di dottorato]

Preview

Text Tesi_Dario_Di_Mauro.pdf Download (3MB) | Preview

Collection description

Nowadays, smart devices populate our environments, providing services and being more and more interactive and user-friendly. However, they usually require a centralised unit that processes all the dialogues to produce an answer. On the other hand, ubiquitous and pervasive solutions are a valid alternative, but it is hard to arrange them in a well-organised environment. In this thesis, I question if a ubiquitous infrastructure can be reactive, flexible and scalable without disadvantaging a uniform environment. Reactivity defines rapid interactions; flexibility concerns both network issues and interactions with users, through customised interfaces; scalability, instead, ensures that the adopted model does not have constrained networks' size. This investigation focuses on Human-Computer Interaction studies, because people without a required technological background will be the final users of the system. I propose a novel distributed model where each node is a device that can independently interact with users through natural interfaces; in addition, nodes collaborate with other similar devices to support people. Nodes' intelligence is limited to their own context. In order to improve the collaboration, devices share partial knowledge and have a common strategy to forward requests they are not able to accept. The resulting network is an Intelligent Environment where the intelligence comes from a composition of connected interactive behaviours. I investigated the best approach to navigate requests, proposing a routing algorithm and considering also security and consistency issues. I contextualised this work in both a smart house and a smart museum. With the devised process, I paid specific attention to professionals involved in the design steps. I identified actors with different roles and needs; in order to meet their requirements, I proposed a designing process, with automated solutions that simplify the implementation of the presented model. The system has been tested in simulated scenarios in order to evaluate all the novel parts. Results showed that the designed model is reactive, flexible and scalable. Furthermore, in order to enhance the final outcome, I characterised design patterns to design the network. Future improvements are oriented to the initialisation of the network, that now requires an expert; In addition, a more complex interaction is under investigation to support users in museum visits.

Downloads

Actions (login required)

View Item