Self-organized Multi-agent System for Robot Deployment in Unknown Environments - Foundations on Natural and Artificial Computation - page 164

Self-organized Multi-agent System for Robot
Deployment in Unknown Environments

A. Canedo-Rodriguez 1 ,R.Iglesias 1 , C.V. Regueiro 2 ,V.Alvarez-Santos 1 ,
and X.M. Pardo 1

1 Department of Electronics & Computer Science, Universidade de Santiago de
Compostela, Santiago de Compostela, Spain
2 Department of Electronics and Systems, Universidade da Coruna, A Coruna, Spain
[e-mail]

Abstract. In this paper, we present a multi-agent system for easy and
quick robot deployment of robots in different environments, without prior
software development and without the need of maps of the environment
where the robot will work. Our system, consisting on a network of intelli-
gent cameras and autonomous robots, is able to manage local interactions
amongst the intelligent cameras spread across the environment, so that
they will self-organize to support robots' navigation and even inform the
robots about the existence of events that require their presence. Our de-
sign is scalable, robust, and flexible to the environment, so that we get
robots able to navigate safely attending events that require their pres-
ence. In this paper, we show the performance of a first implementation
of our system, which was validated in real world experiments, and upon
which we will continue working.

Keywords: Autonomous robots, service robots, sensor networks, intel-
ligent space, multi-agent networks.

1 Introduction

The success of service robots is highly dependant on people's opinion about
the services that they provide. On one hand, robots should show some initia-
tive being able to respond to different events on the environment. On the other
hand, robots must be flexible, reliable and easy to use: robots need to be able
to adapt to different environments and conditions avoiding software or hard-
ware re-design, and they need to ensure continuity of correct service during long
periods of time, minimizing their dependence on specialized support. However,
the ability of the robots to respond to events is limited due to, amongst other
factors, the boundaries of robot's on-board perception. Most of the research that
has been done in robotics has focused on self-contained, stand-alone robots that
act autonomously. This makes the robot to do all the sensing, deliberation and
action selection on board. Nevertheless, the emergence of networked, embedded
systems and the increasing presence of networks in homes and factories, opens
important possibilities towards the integration of robotic technologies with tech-
nologies from the fields of ubiquitous and pervasive computing, sensor networks,