Go to previous article
Go to next article
Return to 2003 Table of Contents
Philippe Truillet, Régis Privat
IRIT UMR CNRS 5505
118, Route de Narbonne, 31062 Toulouse Cedex, France
This article deals with the use of distributed software agents in an open architecture in the "design for all" context. After an introduction to the related concepts, we will show up how the studies on interaction failures can be improved and facilitated in the user-centered design, to achieve more accessible systems using such architecture for rapid-prototyping of effective or wizard of Oz systems. We finally present developed agents already designed in order to build further platforms for our studies.
In this last decade, the new technologies have considerably improved, and are still making progress: mobile phones, powerful laptops, Portable Digital Assistants (PDA), wearable computers, wireless networks... but for whom? In 1995, the concept of User Interface for All (UI4ALL, ) was introduced by Stephanidis. The European Union reformulated it - introducing "design for all". The underlying principle is to ensure the accessibility while considering / respecting the individual needs of the disabled and elderly people.
Until recently, one of the ordinary approach was to realize an a posteriori adaptation of the product / service to take into consideration the users' perceptual and motors characteristics.
But, How to improve systems accessibility? What has to be taken into account while designing interactive systems? Which technological solutions are available? These are some of the questions that developers have to face while designing a system. New approaches have been proposed to solve this problem, such as the concept of the Unified User Interface, a single interface specification to cover all users' classes ().
After a short review of the problems addressed by the researches on accessibility and interaction; we will deal with the distributed software agents approach to facilitate the rapid-prototyping and to get feedback and knowledge in the system design process. We will face this approach to the requirements of the user-centered design process, especially in the phase of prototypes' conception and test, in order to achieve universal access.
2. Accessibility: User / Technology / Context Adequacy
Considering the human-computer interaction has always been a very huge problem. "Accessibility implies the global requirement for access to information by individuals with different abilities, requirements and preferences, in a variety of contexts of use..." (): trying to reach universal accessibility is trying to make the user / technology / context adequacy. Indeed, as soon as one of these "parameters" is changed, an efficient system can become really unusable. In the "design for all" context, it means not only to be accessible in any context of use with the technology implemented and for any given user but also to reach enough efficiency to give user's satisfaction. Otherwise, the system would be more a handicap than a help for the user.
The user-centered design () is an approach that can help in this process of conception of more usable systems, because the design is a process that starts with users and their needs rather than with technology, especially when trying to "design for all". It requires designers to be able to evaluate physical devices, communication and action modalities to see how users employ them in their activities. it also requires to evaluate usability, effectiveness, efficiency and user's satisfaction .
So studying the interaction problems in a user-centered design is a very important step in order to reach accessibility. The facility for a designer to test solutions, make changes or specific adaptations, etc., is a key feature of the design process, and enhancing these capabilities will improve the design step.
3. Enhance Experience Feedback with Distributed Agents
3.1 Agent Concept for a Modular Approach
To address the underlying principles of universal access, the modular approach, which aims at bringing specialized and easily reusable components (processes or encapsulated devices modules), could be benefit by enhancing the experience feedback in giving more information on the interaction needs and ability. Indeed, the modularity allows designing specific-task agents, which could be "spy" to obtain feedback. These agents could be either "physical", interfacing devices and communicating with the system, or "computational". There, we can mention the analogy with the three levels (physical, syntactic and semantic) of interaction specified in the concept of User Interface for All.
3.2 The Rapid-Prototyping Step
In the design process, the rapid prototyping is a very crucial step in order not only to reduce the cost and to test the usability of a system or its technology, but also to improve to knowledge on the user's attitude and behavior, the functionalities of the system (in terms of interaction modalities, dialogue strategies, vocabulary for spoken interaction...). Moreover, designing involves making decisions, pursuing some directions and omitting others. The goal of prototyping is to explore a restricted design space and creating a design that is innovative and makes sense to users in the required context. The feedback experience provided can be used to guide further development but may also be used to transform or reject aspects of the design. The architecture based on such agents would allow rapid prototyping by re-use of existing agents, but it would also allow to study the interaction even if the modalities are not available, by facilitating the realization of wizard of Oz systems.
4. Using Ivy to Connect Applications
Ivy () is a software bus designed at CENA (France) by a research group in Human-Computer Interaction, with the goals of connecting applications written on different toolkits/languages/platforms, while keeping it simple: no server to be launched and supervised, a simplistic API, and a communication model compatible with classical event-based GUI programming. As opposed to other software buses, Ivy does not rely on a central server. An interesting point is that it allows designing customized adapted agents matching a user requirement independently to the system.
Actually, Ivy is mostly a communication convention between processes, implemented through a collection of libraries in several languages like C/C++, Java, Perl,...
5. Our Agents
We developed some agents in order to validate our hypotheses and to test the feasibility of this architecture. We can mention:
Other agents are implemented such as a dialog controller. We use these agents to prototype quickly a service (such as an e-mail application working on several supports - PC and PDA for instance -, and for several kinds of users - blind people, ... -).
In this article, we have shown the interest of a modular architecture based on agents in the design steps. This modularity can be very helpful in rapid-prototyping, allowing not only designing adaptable systems to get knowledge, but also to get feedback on it. This approach, based on a software bus helps us in the "design for all" context, give us further advices and recommendations based on observations and studies of the human-computer interaction. For that, the next step will be to develop agents at the semantic level of the interaction to take the users' perceptual and motor characteristics into account.
Stephanidis, C. Towards User Interfaces for All: Some Critical Issues. Panel Session "User Interfaces for All - Everybody, Everywhere, and Anytime". In Y. Anzai, K. Ogawa & H. Mori (Eds.), Symbiosis of Human and Artifact - Future Computing and Design for Human-Computer Interaction, Proceedings of the 6th International Conference on Human-Computer Interaction (HCI International '95), Tokyo, Japan, 9-14 July 1995, vol. 1, pp. 137-142, Amsterdam: Elsevier, Elsevier Science.
Stephanidis C., Salvendy G., Akoumianakis D., Bevan N., Brewer J., Emiliani P. L., Galet-sas A., Haataja S., Iakovidis I., Jacko J., Jenkins P., Karshmer A., Korn P., Marcus A., Mur-phy H., Stary C., Vanderheiden G., Weber G., Ziegler J., 1998. Towards an Information So-ciety for All: An International R&D Agenda. International Journal of Human-Computer In-teraction, vol. 10(2), pp. 107-134.
Savidis, A., Akoumianakis, D., and Stephanidis C.: The Unified User Interface Design Method. In C. Stephanidis (ed.) User Interfaces for All - Concepts, Methods and Tools. Mahwah, NJ: Lawrence Erlbaum Associates (2001).
Norman, A., D., Draper, W., S. (Eds.), 1986. User-centred system design: New perspectives in Human Computer Interaction, Hillsdale, NJ: LEA.
Go to previous article
Go to next article
Return to 2003 Table of Contents
Return to Table of Proceedings