by Constantine Stephanidis, Stavroula Ntoa, George Margetis and Margherita Antona (FORTH-ICS)

The advent of intelligent environments brings several challenges for the design of universally accessible interactions; however, it also bears novel opportunities. Given the importance of inclusive intelligent environments, we propose a methodology for designing interactions in such environments that are universally accessible.

Universal accessibility refers to the provision of interactive applications and services that are accessible by the broadest possible end-user population, depending not only on their characteristics, but also on the changing nature of human activities, the variety of contexts of use, and the diversity of technological platforms [1]. Simply put, it means that everyone should be able to use an interactive system regardless of their physical, mental, or psychological characteristics, the device they are using, or the conditions under which they are using it. Although this sounds fair, ethical, and obvious, it is certainly not simple, and – apparently – in many cases, this has not been achieved so far.

The concept was initially advocated as the “Design for All” approach, proposing solutions for technological products and services accommodating diversity, and being accessible by people with disabilities [1]. Since then, several advancements have been achieved, be they technical, legal, or societal. As new technologies come into play, new tasks and user goals flesh out, and new contexts of interaction emerge, it is remarkable that universal accessibility not only carries through but actually prospers at the end of the day. For example, consider that not all workstations are equipped with assistive technology solutions for people with disabilities, yet all smartphones come with such features embedded and readily available to consumers.

Our everyday devices have become smarter and interconnected, paving the way for the materialisation of intelligent environments. Even though technological complexity increases, the enhanced capabilities of the new technological environments bring along several benefits for universal accessibility [2]. Intelligent environments are expected to be inherently multimodal, thus accommodating a wide variety of interaction preferences. At the same time, the interconnectivity of devices can be a great asset for achieving personalisation and adaptation to the needs and preferences of each user interacting in the intelligent environment. However, apart from the technological readiness and infrastructure capabilities, there are several challenges that need to be addressed to create universally accessible intelligent environments [3]. In a nutshell, appropriate modelling approaches need to be developed, capturing and correlating user and application characteristics for diverse contexts of use, as well as ready-to-use accessibility solutions, and suitable design tools. Furthermore, as intelligent environments advance toward Artificial Intelligence (AI)-enabled environments, a concrete methodological approach for Human-Centred AI should be framed [4], paying particular attention to how universal accessibility methods, processes, and tools are incorporated.

Our approach to designing universally accessible intelligent environments, applied in the context of the FORTH-ICS Ambient Intelligence Programme [L1] is rooted in the principles of Human-Centred Design, adhering to iterative design organised in four main phases (Figure 1). However, acknowledging that Intelligent Environments inherently employ AI, the proposed approach is enhanced with activities pertaining to the design and development of the AI components as well. Furthermore, for each entailed activity, our approach considers key points that should be addressed for achieving universally accessible solutions.

Figure 1: Methodological approach to designing universally accessible intelligent interactions in intelligent environments.
Figure 1: Methodological approach to designing universally accessible intelligent interactions in intelligent environments.

In the phase of understanding the context of use the overall goal is to comprehend who are the users/inhabitants of the Intelligent Environment, which are the activities that they will realise therein, as well as what are the attributes of the technological, physical, but also the social environment. Important attributes that are specified include user characteristics (e.g. age, physical and cognitive capabilities, psychological attributes, skills, etc.); the envisioned system functionalities and user tasks in the intelligent environment; the available devices, interaction modalities, and assistive technologies; Internet of Things (IoT) and IoT data collected; the auditory, visual, and thermal environment, as well as space and furniture; and also attributes with regard to social interactions in the environment (e.g. social norms, people co-presence, etc.) and application-domain specific goals. The outcome of this activity entails a user model and a context model, which are reusable and extensible.

In the requirements specification phase, appropriate methods are applied to involve in the process representative end-users (e.g., persons with disabilities) in order to create a detailed specification of user requirements in relation to the intended context of use and objectives of the designed system or environment. As such, functional and non-functional requirements are elicited regarding users’ interaction needs and preferences, assistive technologies employed, information needs and preferences, space requirements for approaching and reaching interactive systems, as well as safety and privacy requirements. An update of the devised user and context model is also carried out after this phase.

Informed by the preceding stages, the phase of designing and producing solutions creates the universally accessible solution aiming at the same time to achieve an optimal User Experience. In the context of this phase, the data that will be used to train the system is collected, with particular emphasis on employing datasets that will not lead to exclusion of the target users. Data for AI systems are a recognised factor of exclusion since they often stem from who is thought of as the “average” user, leading to systems that are trained with biased datasets, thus affecting their decision-making. In the same phase, the design of the AI and reasoning process over the devised user-context model takes place, paying attention to universal accessibility aspects and addressing the reactions and information that should be provided by the environment toward anticipating users’ interaction activities. Finally, the user interface design is also a major activity, catering to information and interaction design that is accessible and adheres to the principles of "Design for All" [1].

Last, the phase of evaluation is performed covering aspects of User Experience evaluation, accessibility evaluation, and AI validation, employing automated and semi-automated tools, expert reviews, and, cardinally, user testing. 

The overall process is iterative and scalable; it can be applied to the design of entire Intelligent Environments (such as a smart home), sub-spaces of Intelligent Environments (such as a particular room of a smart home), or particular interactive artefacts (such as a smart table). The work is ongoing to further extend the process in order to indicate appropriate methods and tools for each phase and provide templates for the documentation of each phase, resulting in a pool of use case examples that can be valuable to the research community.


[1] C. Stephanidis, G. Salvendy: “Toward an information society for all: An international research and development agenda”, International Journal of Human-Computer Interaction, 10(2), 107-134, 1998.
[2] P. L. Emiliani, C. Stephanidis: “Universal access to ambient intelligence environments: Opportunities and challenges for people with disabilities”, IBM Systems Journal, 44(3), 605-619, 2005.
[3] G. Margetis, et al.: “Towards accessibility in ambient intelligence environments”, in  F. Paternò, et al. (eds) Ambient Intelligence, AmI 2012; LNCS, vol 7683. Springer, 2021.
[4] G. Margetis, et al. “Human‐Centered Design of Artificial Intelligence”, in G. Salvendy and W. Karwowski (Eds.) Handbook of Human Factors and Ergonomics, 1085-1106, 2021.  

Please contact:
Constantine Stephanidis, FORTH-ICS, Greece
+30 2810 391741, This email address is being protected from spambots. You need JavaScript enabled to view it.

Next issue: July 2023
Special theme:
"Eplainable AI"
Call for the next issue
Image ERCIM News 130
This issue in pdf


Image ERCIM News 130 epub
This issue in ePub format

Get the latest issue to your desktop
RSS Feed
Cookies user preferences
We use cookies to ensure you to get the best experience on our website. If you decline the use of cookies, this website may not function as expected.
Accept all
Decline all
Read more
Tools used to analyze the data to measure the effectiveness of a website and to understand how it works.
Google Analytics
Set of techniques which have for object the commercial strategy and in particular the market study.
DoubleClick/Google Marketing