| Abstract |
Author introduces a framework based on LEGO®
“MINDS STORM®” robots, called “Robo-ant”, that provides
distributed control of large collection of mobile physical robots
in sensor networks. The robots sense and will react to virtual
forces, which are motivated by natural physics laws. This
framework provides an effective basis for self-organization,
fault-tolerance, and three primary factors will distinguish our
framework from others that are related: an emphasis on
minimality, ease of implementation and run-time efficiency.
“Robo-ant” will be implement both in simulation and on a team
of three mobile LEGO© robots and CC1000DK Development kit
(remote transmitter/receiver). Specifics of the robotics bodiment
are presented in the paper.
The focus of author research is to design and build rapidly
deployable, scalable, adaptive, cost-effective, and robust
networks of autonomous distributed LEGO® vehicles. This
combines sensing, computation and networking with mobility,
thereby enabling deployment, and reconfiguration of the multiagent
collective. Author objective is to provide a scientific,
approach to the design and analysis of aggregate sensor systems.
The general purpose for deploying tens to hundreds of such
agents can be summarized as “factor control”. Factor control
means monitoring, detecting, tracking, reporting, and
responding to environmental conditions within a specified
physical region. This is done in a distributed manner by
deploying numerous vehicles, each carrying one or more sensors,
to collect, aggregate, and fuse distributed data into a tactical
assessment. The result is enhanced situational awareness and
the potential for rapid and appropriate response. Author goal is
to design fully automated, coordinated, multi-agent sensor
systems. An agent’s sensors perceive theworld, including other
agents, and an agent’s effectors make changes to that agent
and/or the world, including other agents. It is assumed that
agents can only sense and affect nearby agents; thus, a key
challenge has been to design “local” control rules. Not only do
author want the desired global behavior to emerge from the
local interaction between agents (self-organization), but also
require fault-tolerance, that is, the global behavior degrades
very gradually if individual agents are damaged. Self-repair is
also desirable, in the event of damage. Self-organization, faulttolerance,
and self-repair are precisely those principles
exhibited by natural physical systems. Thus, many answers to
the problems of distributed control can be found in the natural
laws of physics. This paper presents a framework, called “Roboant”.
Although the forces will be virtual, agents act as if they
were real. Thus the agent’s sensors must see enough to allow it
to compute the force to which it is reacting. The agent’s effectors must allow it to respond to this perceived force. Author
see two potential advantages to this approach. First, in the real
physical world, collections of small entities yield surprisingly
complex behavior from very simple interactions between the
entities. Thus there is a precedent for believing that complex
control is achievable through simple local interactions. This is
required for very small agents, since their sensors and effectors
will necessarily be primitive. Second, since the approach is
largely independent of the size and number of agents, the results
scale well to larger agents and larger sets of agents. Three
primary emphases distinguish the “Robo-ant” framework from
others that are related: minimality, ease of implementation, and
runtime efficiency.
First, robots formations will be achieved with a minimal set
of sensors and sensor information. The rationale for this
emphasis is that it will: (1) reduce overall vehicle cost, (2)
enable physical embodiment with small agents, and (3) increase
vehicle stealthiness if sensing is active. Second, the paper
presents the toretical results that translate directly into
practical advice on how to set system parameters for desired
swarm performance. This makes the robotic implementation
straightforward. |
