![](bg2.png)
cooperative control of networked collaborating mobile robots with distributed resources
such as sensors, computing power and communications [2, 10, 8].
Consider the problem where of a group of mobile robots has been dispersed in a
given area, and that it is required to gather the robots in the vicinity of a given point.
For example, consider the case where a group of robots has to perform a mission in
a remote area, where they have landed by parachutes. The robots are now randomly
scattered in a wide area and need to be gathered into a much smaller area in the vicinity
of a designated location before starting their mission. The specific task now is for the
robots to collectively move towards the gathering point. We consider swarms on the
order of 200 robots dispersed in a two-dimensional domain on the order of 1 km by 1
km.
It is assumed that each autonomous robot is equipped with a compass and is capable
of moving in a given azimuthal direction for a given distance. Each robot has a low level
control and navigation system that can detect its location at all times and guide it from
one point in the domain to the next at the right speed and orientation. It is also assumed
that each autonomous robot is equipped with a collision and obstacle avoidance control
system for preventing collisions with other robots and obstacles. The robots network
architecture consists of a leader robot acting as a server and communicating with the
other robots as clients.
The robot swarm cooperative control method is described in the next section. Each
robot has a microprocessor computing device on board capable of running the robot
swarm algorithm. We propose to use this paradigm algorithm as a top level discrete
event controller for the cooperative control of the swarm. Each robot sends the best so-
lution found at any given time to the leader or other central processing station through
its communication channel. The leader in turn computes the global best solution and
transmits the result as a control signal to the network. The Robot Swarm Optimization
(RSO) is a stochastic population based method that belongs to the class of biologically
inspired algorithms. It is based on the paradigm of a swarm of insects performing a col-
laborative task such as ants or bees foraging for food using chemical or some other type
of communication, see for example [1] and [3]. The swarm intelligence method was
originally developed by [4] and later described in great detail in [5]. An overview of the
method as extensively applied to various function optimization problems of increasing
difficulty has recently been presented by [7]. Here the PSO method is and adapted for
use as a top level discrete event cooperative control method for a swarm of autonomous
robots.
In the next section we develop the robot swarm algorithm with communication noise
and we explain how it can be applied to solve the swarm gathering problem. In section
3, results of simulations are described for a swarm of 200 robots, gathering in a noisy
environment. We show that the robots trajectories follow Levy flights and compute the
probability distribution for the flights lengths.
2 Cooperative Control of the Robot Swarm
In developing the robot swarm cooperative control method, we incorporate physical
effects or constraints in order to implement the search method by actual mobile robots
4