Here we will explain how things work in a family of birds. In general, some birds of the family will collect food, known as the food gatherers; some birds of the family will collect the sticks needed to strengthen or build a new nest; some birds of the family (typically the elder, more experienced ones) will teach baby birds how to fly; and other birds can find hay to cushion the nest. But as any such system, the bird swarm system will also suffer from conflicts, which are dealt with at the end.

Communication will mainly occur in ways such as: birds communicating by chirping when they are close to each other, or birds wanting to communicate on longer distances by certain signals of their own. Also, family members may be informed that other members have explored a particular area for instance, by leaving markings.

Now that we have analyzed how a family of birds co-operates and coordinates, we can outline its implementation using a swarm system:

• All birds of the family will be individual computing entities (agents).
• The environment in which these agents will be situated is a natural setting where food, twigs and hay are resources that will be present in abundance
• All bird agents will act autonomously where they can move around, collecting resources to make a comfortable home.

In this animation, we can see that a heterogeneous swarm system has been simulated. Tom has been assigned the task of collecting twigs for the nest. On the hunt for twigs, he finds food. He communicates via chirpping informing the bird (Chip) that was suppose to collect food to now collect twigs. We say that they have swapped roles, both birds have the capablity of collecting food and twigs.

Now follows a detailed description of the various different aspects of a multi-agent system, entailing the practical application of what we have briefed in general.

As in any family, birds will require interaction to make sure they work towards a comfortable home.
As mentioned earlier in the site, there are various forms of interaction. Here we mention a few.

• Simple collaboration
  
  This type of interaction may be required when there are many tasks to be done, but not sufficient birds to them. At this point, the bird agents will require interaction in order to collaborate tasks and a bird may be assigned more than one task. For example, to search for food, as well as teach their flock how to fly.


• Obstruction
  
  This happens when the task of one bird agent is hindered by either a limited resource, or the environment situation. For example, the bird agents all leave for hunting the different things that are required for the nest. Now assume that the size of the nest only allows one bird agent at a time. At this point, interaction will be required in order to find out which bird needs to pass through first.

In the natural environment, birds communicate by gestures and chirping. In a Multi-Agent System, they communicate by passing messages to each other. Messages can take the format of the KQML style, and be incorporated by using a specific common language, that allows effective transmission. Speech acts could also be adhered to when bird agents communicate. Mobility of the robots in the system will be made efficient by placing sensors into them so that two birds do not bump into one another.

There are several situations where a conflict can arise within the bird multi-agent system. For example, two birds (Bird A, Bird B) are teaching their flock how to fly in two different ways. The two birds fly around in different directions spreading their knowledge on how to fly to every baby bird they come across. If Bird A teaches a particular bird and then, Bird B comes across the same baby bird, the bird in question is taught how to fly again. The two birds might have different teaching approaches to the lesson, and this may confuse the baby bird.

Relating this to the agent system, the birds can be depicted as agents. Bird A can have a different knowledge base in comparison to Bird B, giving rise to the different teaching styles. When the bird being taught is confused, the agents (Bird A and Bird B) will be informed. By this we mean that the bird will interact with both these birds, informing them of its situation and about the conflict that has arisen. Now, birds A and B will require a one-to-one negotiation, in order to resolve the conflict. It is resolved once both birds come to a mutual agreement on the technique of flying.