Année 2014 - Numéro spécial Éditeurs : Sybille Caffiau & Celia Martinie
Multi-user multimodal interactive systems involve multiple users who can use multiple interaction modalities. Multi-user multimodal systems are becoming more prevalent, especially systems based on large shared multi-touch surfaces or video game centers such as Wii or Xbox. In this article we address the description of the tasks with such interactive systems. We review existing notations for the description of tasks with a multi-user multimodal interactive system and focus particularly on tree-based notations. For elementary tasks (e.g. actions), we also consider the notations that describe multimodal interaction. The contribution is then a comparison of existing notations based on a set of organized concepts. While some concepts are general to any notation, other concepts are specific to human-computer interaction, or to multi-user interaction and finally to multimodal interaction.
Existing task models have often been used in the context of graphic systems. In this paper, we propose to use the task model at runtime to monitor user actions, to verify that he/she has not made a mistake when performing his/her actions and to give him/her help when necessary. In particular, we present, as a first contribution, a task model specific to interactions in ambient environments. This model enables to assign dynamic characteristics to each task thereby allowing to a supervision system to assign states to tasks at runtime based on the information exchanged with the environment (start of a task, end of a task, preconditions states...). Our second contribution is a monitoring and support system that exploits our task model. More precisely we specify the intervention strategy of our system in order to guide the user. We present then an illustration of our system through the execution of a scenario on our simulator. This simulation shows how the interactions with the task model at runtime allow us to produce a dynamic system that takes into consideration the context and provides assistance to the users while carrying out their daily tasks. Finally, we end with a conclusion and perspectives of our approach.
La description de l’activité humain-système par des modèles de tâches existe depuis plusieurs années. Après l’époque papier-crayon, les outils d’édition des modèles de tâches ont permis la conception et l’archivage des modèles selon une notation rigoureuse, et la vérification de leur cohérence. Mais la compréhension de la dynamique est restée affaire de spécialiste jusqu’à l’apparition des simulateurs de modèles. Une simulation permet d’appréhender les enchaînements réels de tâches - décrits implicitement par les opérateurs temporels - et de valider les scénarios ainsi réalisés. Cet article décrit et compare les simulateurs actuellement disponibles et maintenus, et explique leurs différents usages en fonction des buts et niveaux d’expertise des utilisateurs. De nouvelles perspectives d’évolutions de ces modèles et outils sont alors définies dans le but d’améliorer leur sémantique.