HUManAID-Sens

Motivation

Previous research has proven the impact that the human factor has on intelligent environments, where there is a challenging gap between what is provided by an intelligent system and what is really needed to support both full autonomous functioning within the given context and better catering for the user needs. However, despite the clear benefits, personalized systems are still scarce in real–world applications, mainly because of the level of difficulty associated with providing this type of support.

Project target

In HUManAID project, we try to fill this gap by building a framework that supports the development of adaptive systems, adopting a user–centered design. To this end, we gather the wide expertise of the 4 subgroups that conform the project in different application areas, to bring it together under the construction of a Digital and Ecological Transition based on Automation (DETA) framework that facilitates the development of systems that behave according to the evolving needs of users and their traits. To tackle this gap and progress on the level of autonomy and performance of these intelligent systems, this project focuses on developing a common ground of digital user–centric intelligent technologies that combine inter–subject and intra–subject approaches within highly sensed scenarios that consider ecological needs. The purpose here is to understand and detect user and system needs in an innovative collaborative way user–system/system–user using the minimum electronic and computational resources so that the user increases his/her performance and satisfaction along with the system, which in turn increases its level of autonomous functioning and adequacy to cater for the user and the context changing needs within the given situation.

In this way, through the proposed DETA framework, the project covers several key issues related to the digital and ecological transition:

1. Go beyond ad–hoc personalization in domain independent interactive systems so both users’ performance and trust are improved;
2. Carefully define how to develop the sensing and acting infrastructure so that ecological considerations, related to low intrusiveness approaches, are taken into account and thus make the systems really applicable in real world environments;
3. Address the gap between building general models, that cannot support personalization, and individual models that are not able to generalize, using semi–supervised learning techniques, which also contribute to produce digital solutions that are ready to be used in the real world, and not just in the laboratory, as typically happens with adaptive systems; 
4. Develop procedures and tools that ensures fairness, accountability, transparency, gender considerations and inclusion supporting privacy and ethical standards throughout project developments and dissemination activities.