ARCoS - Autonomous Mobile Robot Control System
Intelligent mobile robots can be seen as autonomous agents capable of flexible
action in real world environments.
The kernel of an autonomous robot is its control system whose design is
a very challenging field (see also "Robots
as responsible Agents" by Eugenio Oliveira, 1997).
ARCoS - Autonomous Mobile Robot Control
System aims to transform a mobile platform in a real Autonomous
Robot Agent- ARA,
able to deal with dynamic environmemts.
The proposed ARCoS architecture is a Hybrid Architecture integrating
both the reactive and deliberative paradigms. It has an heavy influence
from the Behavior-based Approaches and it is based on the Multi-agent
Methodology for software developing.
The ARCoS kernel is the reactive agents community, where each
one of the community's member, the
Reactive Agent RA, is responsible for one navigation competence.
Each RA is also capable of cooperating with the other RAs combining efforts
to execute a more complex task. Whenever previous knowledge, about the
environment where the task is taking place, already exists ARCoS also includes
another kind of agents, the deliberative agents, whose role is to advise
the reactive community, thus allowing the mobile robot to exhibit what
we call an oriented reactivity.
Due to uncertianty and imprecision always present in real world
environments, we decide to use the Fuzzy Logic Approach to implement
the reactive agent competencies as well as the cooperation strategies.
In order to allow the mobile robot to learn and adapt to new situiations,
each reactive agent is able to learn from a Neural Network with
a simple architecture and a kind of reinforcement learning mechanism.
On the pratical side, our mobile robot platform is a Robuter II, from
RobotSoft, evolving in an 3D environment with obstacles.
Maria da Conceicao
Last Updated: 99/02/01