Information Technology Reference
In-Depth Information
- a few standard user agents: one is enough in the event where a single user
travels the length of the underground passage; for the experiment with three users,
we have implemented three of these agents;
- five presenter agents, each responsible for one of the screens.
During the practical realization of the experiment, we used five notebook
computers to represent the five screens, all the while keeping the same architecture.
However, in order to avoid problems related to network connections between
computers
8
, we preferred to duplicate the informative agent and the user agents
on
each of the five notebook computers
.
In summary, each computer therefore contained:
- a copy of the informative agent;
- a copy of all the user agents linked to the infrared localization system;
- the corresponding presenter agent.
As a general rule, such a duplication of agents could pose a problem, as different
instances
of the same agent could be found in different states, which would be
incoherent.
However, in the very restricted framework of our experimentation, this did not
disrupt the global functioning of the system, as the user agents implemented did not
have any particular
states
.
10.9. Conclusions and perspectives
We have introduced a conceptual model for the presentation of multimodal
information to mobile users, the natural application of which is in train stations and
airports. This model is accompanied by algorithms for the choice of modalities
according to the abilities of the interactive devices and the users. This model and its
algorithms were implemented in the PRIAM platform, which enabled us to conduct
experiments in pseudo-real conditions. These experiments have shown the benefit of
carrying out a selection as to what information is presented in public places,
depending on the people situated in front of the presentation devices.
The people who participated in our experiments often told us that they had been
unsettled by modifications in the content of the screen, which were too frequent.
8 Even though at the beginning of this chapter we formulated a hypothesis stating that all the
entities would be realized by a wireless network, we preferred not to make our experiments
dependent on possible problems linked to the network.
