TASK MANAGEMENT AND ITINERARY PLANNING
An Integrated View based on Multi-Agent Systems
Fábio L. Correia, Rui F. S. Amaro and Rosaldo J. F. Rossetti
Artificial Intelligence and Computer Science Laboratory (LIACC), Department of Informatics Engineering (DEI)
Faculty of Engineering, University of Porto (FEUP), Rua Dr. Roberto Frias, S/N, 4200-465 Porto, Portugal
Keywords: Task management and planning, Activity-based systems, Dynamic route guidance, Ubiquitous computing,
Environment intelligence, Multi-agent systems.
Abstract: The main objective of this paper is to describe the framework of an agent-based agenda manager. The
technology herein presented is intended to be able to assist a user in his/hers every day’s life, supporting all
aspects of activities to be carried out in different places and times. Differently from other tools with the
same ability, our agenda agent goes beyond the simple task of managing activities and their common
attributes, such as date, time and place at which an activity is to be performed. Profiting from all potentials
offered by mobile communication and portable devices, activities must now be assisted throughout their
whole lifecycle, meaning the user will be able to optimize his/hers daily agenda including journeys he/she
must make between places of two consecutive activities. This paper reports on the first steps towards the
specification of the whole multi-agent architecture for an agenda management system, accounting for real-
time and geographical distribution constraints that are inherent in this kind of systems.
1 INTRODUCTION
In today’s hectic world, professionals in various
areas of activity have to multiply themselves into
many individuals (as if that was possible) in order to
manage their professional and personal lives.
Indeed, life today is extremely demanding and
professional, as well as personal activities are
usually meshed up and overlapped, which increases
conflicts of interests that must be minimized when
tasks and activities are being planned. Our goal is to
conceive a system offering the adequate means to
blend their personal and professional agendas,
accounting for their geographical distribution and
allowing fast and dynamic adaptation as the day
evolves.
Thinking of different potentials of such a system,
it is easy to identify ways in which it could
contribute to the alleviation of the stress associated
to dealing with several compromises simultaneously.
For instance, many companies depend on delivering
or receiving products and this depends on one’s
ability to transport something somewhere as quickly
and efficiently as possible. Current technology
already offer us devices that make everyone’s tasks
easier by calculating the best routes according to
traffic information, thus saving time that can be used
in the accomplishment of other tasks.
Companies also have a huge to-do list which has
to be performed throughout the day, week or month,
accounting for the various constraints each activity
will impose. Managing such individual
characteristics of every single task on an integrated
and optimal basis represents a time-consuming task
anyone would like to avoid.
Some tools, however, are frequently applied as
an attempt to better manage activities and their
constraints. Not surprisingly, many people still find
great use in old-fashion agendas, where tasks are
listed and contacts are maintained. Certainly new
technology has already proven to be a fascinating
way to turn these agendas into a more productive
and efficient way to manage duties. They have been
incorporated into our desktop environments,
electronic mail applications and more recently have
they proliferated among portable devices such as
PDA and mobile phones. These applications have
the ability to organize one’s day, week or month, to
set up an alarm on/off to warn us of a certain event,
to set tasks’ relative importance, or even to integrate
information among different thematic agendas into
an integrated environment.
361
L. Correia F., F. S. Amaro R. and J. F. Rossetti R. (2009).
TASK MANAGEMENT AND ITINERARY PLANNING - An Integrated View based on Multi-Agent Systems .
In Proceedings of the International Conference on Agents and Artificial Intelligence, pages 361-364
DOI: 10.5220/0001805003610364
Copyright
c
 SciTePress

However, all afore mentioned gadgets require a
frequent interaction with the user that needs to
explicitly set up all attributes and task information
so as to make them work properly on his/hers own
benefit. Another factor that increases even more the
complexity of such a scenario is the fact that many
tasks are generally geographically distributed. This
implies the user needs to plan journeys between
activities that are to be performed in different places.
In tricky situations when no time is left for the user
to plan an itinerary in advance, he/she is at risk of
arriving late at a certain activity. Thus, managing
activities is much more than just avoiding time
conflicts or ordering them in an optimized way. It
will involve planning itineraries between tasks as
well.
Our goal in this work is to devise a system
capable of integrating all aspects related to task
management and itinerary planning as a way to
improve daily agendas, taking into account user’s
preferences and performance measures. We will rely
on the concept of autonomous agents and multi-
agent systems to provide us with the necessary
architecture to implement such a system.
Technologies such as GPS and route guidance
systems, as well as contacts and calendar tools will
be used in the conceptualization of our framework,
which is expected to contribute for a reduced stress
and better quality of life.
2 RELATED TECHNOLOGIES
Let us take a look at the most common structure of
an agenda management application nowadays. The
user has to input the task’s or the event’s
characteristics and has to manually format them. For
example, if we use applications such as Microsoft
Outlook or Mozilla Sunbird, we have to introduce
the duration of tasks, their frequency (if it is a daily,
a weekly or monthly task), the location, if we want a
reminder or not, when this reminder should be set
on/off and so forth. Also, once the task or event is
completed it is up to the user to check the calendar
entry as completed or if one was not able to
complete a certain task, again, it is up to the user to
postpone the calendar entry.
These applications have the possibility to be
synchronized with a cell phone and/or PDA which
give the user a certain mobility freedom. However,
they are not clever enough so as to self adjust
according to the user’s needs.
The scenario given above has been addressed in
several research works. For instance, Berry et al.
(2006) devised the PTIME agent that is able to learn
from its interaction with the user in order to improve
its agenda. The multi-agent approach has also been
applied to this kind of domain with relative success,
as in (Modi et al., 2005). In most of these works,
however, authors are focused on scheduling rather
than on managing the whole life-cycle of tasks,
including their geospatial constraints as well. Our
approach then differs from the others as we focus
precisely on the integration of tasks scheduling and
their in-between route planning.
Imagine someone who has a family with children
and works somewhere. This person wakes up early
in the morning and has to plan the day. He/she has to
leave his/hers children at school, go to work, visit
some clients, meet other companies’ representatives,
pick up his/hers children at school and fetch some
house supplies on the way home. Some of these
tasks are performed on a daily basis, like taking the
children to school and picking them up, for instance.
But others are sporadic, such as stopping to buy
some groceries on the way home. Some are
predicable, such as having to meet some clients but
others are not, as having an urgent meeting at the
company.
Bearing this picture in mind, consider that this
person has his/hers calendar with everything he/she
has to do during the day, week and month ordered in
a certain preferable way. While arriving at the
office, he/she finds out that a meeting has been
postponed but he/she has to go somewhere to meet
another client. By introducing something like “meet
client X at certain place A,” an intelligent calendar
manager could connect to the Internet. Using an
application such as Google Maps and receiving
his/hers GPS coordinates, the system computes a
route from his/hers current workplace to the exact
location where the meeting is going to take place.
Arriving at the destination, the calendar manager
knows the user is a bit early and recognizes some
grocery shop in the surrounding area and suggests
he/she might fetch the groceries as an attempt at
anticipating a task which is planned for later on.
Nowadays, we have almost every single
technological aspect required to have such a service,
meaning we have GPS devices through which one
can see points-of-interest (POI), tools that can
re-plan routes according to real-time traffic
information, such as TomTom (www.tomtom.com),
NDrive (www.ndriveweb.com) and Pioneer’s
(www.pioneer.eu/eur/page/products/NavGate/landin
g.html) navigation systems, to mention few.
Unfortunately, none of the systems mentioned
above provides a really reliable managing service
ICAART 2009 - International Conference on Agents and Artificial Intelligence
362

which is able to establish a bridge between the
geographical planning (itinerary selection) and the
time scheduling (calendar management) with such
intelligence and awareness.
The objective of the work herein described is to
incorporate in an interactive service all these
aspects. It is our intention to promote the task
management from a static view to a dynamic
perspective where a lot of environment variables
would play an important role. A service capable of
rearranging one’s tasks efficiently, taking into
consideration not only the task’s time constraints but
also the user’s relative position, the commuting
times between tasks’ places, the user’s shop
preferences and so on. Considering these aspects the
tool we are describing would optimize the user’s
time and geographical management, consequently,
maximizing his/hers efficiency.
Taking this dynamic view at task management,
we are going to be able to create a new service
which will provide the professionals of the future
with a time management aid which will be crucial in
today’s demandingly growing world.
3 THE AGENT-BASED
ARQUITECTURE
We based our approach on the autonomous agents
and multi-agent systems (MAS) metaphor, which
represents a natural way to model our application
domain. This approach is not actually new, as other
authors have also investigated the potentials of MAS
applied to this field (Modi et al., 2005). The novelty
in our approach, nonetheless, lies on the integration
of tasks management and itinerary planning as part
of the activities handled by the same multi-agent
systems that, to the best of our knowledge, has not
been considered in other approaches. The multi-
agent concept fits our idealized service because we
would have three subsystems which would
separately play an important role but still
communicate with each other on a cooperative basis.
The three multi-agent subsystems identified are i) a
server oriented system, ii) a pc-client system, and iii)
a mobile client system (see Figure 1).
The server oriented system encompasses an
agent to manage tasks such as establishing the
connection between the mobile-client and the pc-
client whenever the mobile-client is out of the
normal short wireless communications range. It also
includes a synchronization agent in order to
synchronize possible changes between the mobile
and the pc calendars accordingly.
Figure 1: Multi-agent subsystems overview.
In the pc-client system, a calendar management
agent, a synchronization agent, and a route planner
will share their knowledge on user’s preferences in a
common database. This database stores information
such as the commuting times, in order to provide the
route planner with some information on user trip
performances, for instance. The calendar
management agent supports most user tasks and is
able to advise the user on how to plan his/hers day,
week or month. This system also incorporates a
traffic-monitoring agent to take advantage of most
up-to-date and real-time information on traffic
conditions. The synchronization agent is able to send
information to the server, which would then be read
and used by the mobile-client whenever the user is
out of the office and dynamically needs on-route
information about his/hers agenda.
The mobile-client system also includes a
communication agent, capable of autonomously
deciding whether a connection to the server or pc-
client is needed and will manage the necessary
resources to accomplish so, including WiFi, GPRS,
3G, Bluetooth or wired communication capabilities
whenever they are available for use. This agent
should know then when to use the server and what
task should not be locally processed, thus
minimizing performance loss if avoiding it is not at
all possible. Also, the mobile client system
encompasses a GPS agent in order to monitor user’s
position. This agent interacts with calendar
management agent, whenever possible (this
communication is managed by the communication
agent), to define the next destination for the user.
TASK MANAGEMENT AND ITINERARY PLANNING - An Integrated View based on Multi-Agent Systems
363

The system also includes a calendar agent, similar to
the one included in the pc-client, where the user is
able to introduce sudden and/or unexpected tasks
making the necessary adjustments autonomously.
An intelligent SMS agent is also present, meaning
one might receive a SMS informing about a meeting
or a client visit being automatically introduced in the
calendar and making any necessary adjustments.
Accounting for agents’ ability to interpret natural
language, such as the one proposed in (Modi et al.,
2005), the system could autonomously create a new
calendar entry simply by recognizing it within the
body of an email received by the user.
The portable device should do as few processing
jobs as possible in order to maximize efficiency and
autonomy. In order to do so, some tasks are only
performed by agents when deemed really necessary.
For example, when an email arrives with a new task,
the pc-client system computes the necessary
adjustments and then sends a message to the mobile
device letting it know it should connect to the server
so as to receive some new important information. If
the new task is considered to be secondary, the pc-
system will await a wired or Bluetooth connection
next time the user is in the office. This could happen
with a lot of new entries, meaning that whenever a
new task is assigned to the mobile calendar, it
should be clever enough to know if there would be
major rearrangements letting the pc-client handle it.
All of the above mentioned subsystems are to
interact and cooperate with each other in a certain
way so that the user can have, in a single system, the
functionalities of a full calendar manager, a full GPS
navigation system, and the ability to combine both to
maximize their potentials.
4 CONCLUSIONS
Mobile communication and computing give rise to a
wide range of new applications. Coupling traditional
calendar and task management features encountered
in most desktop applications with the ability to
wisely plan journeys between tasks to be performed
in different locations seems to be a very promising
application of such technologies. This is certainly
one step forward towards a world where frontiers
between human users and technology would no
longer be evident.
We have conceived a multi-agent based task
management system formed by three smaller multi-
agent subsystems. These subsystems are intended to
wisely cooperate with each other profiting from
today’s abundant information resources. The system
provides user with a device functioning as his/hers
own personal assistant, aware of where he/she is,
communicating autonomously with his/hers calendar
and computing the best way to go through the day
wasting as less time as possible.
The implementation of this system is in its early
stages and is based on the integration of many
different technologies, from distributed systems, to
autonomous agents and geospatial analysis.
Nonetheless, all applications expected to interact
with the user should be user-friendly as suggested by
Wei and Rudnicky (2000).
Many issues arise from this complex scenario,
which are related to the autonomy of mobile agents,
communication protocols and implementation
viability, as identified in (Zhang et al., 2007). Also,
nowadays users are quite tied to service fees in order
to send and receive emails and SMS, as well as to
keep long lasting Intenet connections in their
handset devices. So, addressing cost issues is
another concern in this development too.
The very next steps in this research will include
the definition, adaptation and/or implementation of a
middleware to allow the above mentioned agents to
interact with each other and with the user. The user
would not have to worry about setting his/hers GPS
device or rearranging his/hers week on separate
applications, but every related task could be carried
out on one single application.
REFERENCES
Modi, P.J., M. Veloso, S.F. Smith, J. Oh (2005) CMRadar:
A Personal Assistant Agent for Calendar Management.
Agent-Oriented Information Systems II. LNCS 3508.
p.169-181.
Zhang, Y., B. Hull, H. Balakrishnan, S. Madden (2007)
ICEDB: Intermittently-connected continuous query
processing. In Proc. of the IEEE 23rd International
Conference on Data Engineering. p.166-175.
Wei, X., A.I. Rudnicky (2000) Task-based dialog
management using an agenda. In Proc. Of the
ANLP/NAACL Workshop on Conversational systems.
v.3, p.42-47.
Berry, P., K. Conley, M. Gervasio, B. Peintner, T. Uribe,
N. Yorke-Smith (2006). In Proc. of the fifth
international joint conference on Autonomous agents
and multiagent systems. p.1564-1571.
ICAART 2009 - International Conference on Agents and Artificial Intelligence
364