The i-Walker: An Intelligent Pedestrian Mobility Aid - Computational Intelligence in Healthcare: Advanced Methodologies - page 108

known environment, so the va has available a semantic map of the world with tags (see
Fig. 4). The SHARE-it agent layer relies on an Ontology for knowledge representation,
which includes concepts related to space and contained objects among other facets of
the SHARE-it system reality. All this relevant semantic information can be integrated
with the topological maps in order to obtain a semantic map that the agents can use
for higher level planning tasks. This high level information if used for Route planning
service, where the user selects a place where it wants to go using room images and the
va shows him instructions on how to reach there; although available, autonomous nav-
igation is not recommended by doctors, as is does not foster user autonomy and could
lead to capacity loses. The va also detects falls in the i-Walker , generating an alarm
to the pa to request help to the caregivers just in case the user is hurt or can not get
up. Like the ea ,the va performs component monitoring and when some of the system
part stop working, generates a SMS Technical alarm that is sent to the administrators
to take care of it. The users navigation is also monitored, and depending on the users
health status and profile, the agent will make decisions on the degree of control the user
will have on navigation. The va can modulate the amount of force the user has to do
on each handlebar or the amount of help he gets on them. Monitoring the user forces
behaviour and merging that information with his updated disability profile, the agent
can modulate the exervice or help granted to the user on every situation.
The Home Agent ( ha ) performs a global monitoring task, and stores all the media
information needed for the personalized interface of the users or the media files used to
build up the dynamic tutorials. The ha also stores the agenda or the user profiles that
are sent to the pa regularly. The Caregiver Agent ( ca ) is a information frontend that
allows the caregivers to watch whats going on with the elder user, if any alarms have
been raised, which activities have been performed along the day or to access the user
agenda and modify its entries.
Multi-agent systems have both the flexibility and the cognitive capabilities required
in order to be able to support the needs of persons with different disability profiles
and to complement the autonomy of the people with special needs in an adaptative
way through the time. In some cases the disability is a consequence of a pathology,
that may improve with some time and rehabilitation. An excess of support or lack of
flexibility in the support can make this process more difficult, on the other hand an
assistance adaptive to the daily state of the patient may be helpful in the rehabilitation
process. Agents have proven their capacity of being flexible, adaptable and learn from
experiences with the proper design.
Some patients may dislike an autonomous navigation system, or choosing among
a set of maneuvers, they may prefer driving by themselves, to feel autonomous and in
charge of the situation at all times. An intelligent agent with the necessary knowledge of
a user's profile can supervise user's navigation and take part in some driving maneuvers,
in a transparent way, in case the user needs some support ( e.g. help crossing doorways,
refining turning maneuvers, help keeping stable cruise navigation, ...).
In order to make this possible the user's agent must have to have deep knowledge of
the user's disability profile and historical data about his/her driving behaviour, merge all
this knowledge and translate it in control support and a set of assistive services. All this
knowledge and information must be updated dynamically, since the user can progress