Information Technology Reference
In-Depth Information
way similar to the use of pheromones in the ant systems. However, this solution has a
drawback when a few robots (and agents) for emptying bins, are used in the system.
In this case the attracting of a full bin may attract a few agents and consequently the
robots, which gather around the bin without any purpose, whereas only one robot is
needed (one agent in the cyberspace).
This drawback may be overcome by taking advantage of the capability to produce
the smell not only by full bins but also by the agents (or robots) circulating in the
environment. The difference in smell is that the smell of one agent is repulsive to
another agent. Spreading such smell in the environment prevents agents (and robots)
from gathering around one full bin.
Therefore, we may consider that each agent produces its own characteristic
smell. Consequently, in the environment different types of smells with the following
properties are spread:
Smell produced by a given agent is repulsive to others,
Smell does not affect the agent which produces it,
Full bins placed in the environment also produce smell which attracts all
agents.
The agent after getting to the node of the graph has the capability to check the
concentration of individual smell related to the edges of a given node, and on the
basis of the result of these observations makes a decision along which edge it should
continue its route. It takes into account that:
the smell of full bins attracts,
the smells of other agents are repulsive.
If there is a choice between two or more edges with identical concentration of
smell, the right edge is chosen randomly.
By developing the example presented above, we may consider different ways of
producing smells as well as different ways of decision-making on the basis of smells
appearing in the environment.
This system may be an example of the application of the robot management in
real-space with the use of simulation of the real environment in cyberspace, with the
application of the concept of the agent.
Obviously, having some information about full bins and the location of robots
which empty these bins, we may consider other methods (algorithms) of the robot
management. The considerations we present are an example of the agent approach
to the solution to a problem rather than the optimal solution to the task.
5.1.4 The Agent System Managing Task Distribution Among
Mobile Robots
The agent system may be used for the management of task distribution among
mobile robots. In the exemplary system, the agent system is responsible for such
