Behavioral Animation - Computer Animation: Algorithms and Techniques - page 341

11.1 Primitive behaviors
Primitive behavior , for purposes of this discussion, is characterized by being immediately reactive to
the sensed environment as opposed to any temporally extended reasoning process. The environment is
sensed and an immediate reactive behavior is made to the conditions extracted from the sensory data.
There is little reasoning, no memory, and no planning. Usually, a simple rule-based system is involved.
While a limited number of internal states might be modeled, they control basic urges such as hunger and
flight-from-danger. Flocking and prey-predator behavior are the primary examples of such primitive
behavior that is of a direct cause-and-effect form.
11.1.1 Flocking behavior
Flocking can be characterized as having amoderate number of members (relative to particle systems and
autonomous behavior), each of which is controlled by a relatively simple set of rules that operate locally.
The members exhibit limited intelligence and are governed by basic physics. The physics-based model-
ing typically includes collision response, gravity, and drag. As compared to particle systems, there are
fewer elements and some interactions with nearby neighbors are modeled. In addition, members' behav-
ior usually models some limited intelligence as opposed to being strictly physics based.
Flocking is one of the lowest forms of behavioral modeling. Members only have the most primitive
intelligence that tells them how to be a member of a flock. From these local rules of the individual
members, a global flocking behavior can emerge. While flocks typically consist of uniformly modeled
members, prey-predator behavior can result from mixing two competing types of mobile agents.
The flocking behavior manifests itself as a goal-directed body, able to split into sections and re-
form, creating organized patterns of flock members that can perform coordinated maneuvers. For pur-
poses of this discussion, the birds-in-flight analogy will be used, although, in general, any collection of
participants that exhibits this kind of group behavior falls under the label of “flocking.” For example,
flocking behavior is often used to control herds of animals moving over terrain. Of course, in this case,
their motion is limited to the surface of a two-dimensional manifold. To use the flock analogy but
acknowledge that it refers to a more general concept, Reynolds uses the term boid to refer to a member
of the generalized flock. Much of this discussion is taken from his seminal paper [ 34 ].
There are twomain urges at work in themembers of a flock: avoiding collision and staying part of the
flock . These are competing tendencies and must be balanced, with collision avoidance taking
precedence.
Avoiding collisions is relative to other members of the flock as well as other obstacles in the envi-
ronment. Avoiding collision with other members in the flock means that some spacing must be main-
tained between the members even though all members are usually in motion and that motion usually
has some element of randomness associated with it to break up unnatural-looking regularity. However,
because the objects, as members of a flock, are typically heading in the same direction, the relative
motion between flock members is usually small. This facilitates maintaining spacing between flock
members and, therefore, collision avoidance among members. The objective of collision avoidance
should be that resulting adjustments bring about small and smooth movements.
Staying part of the flock has to do with each member trying to be just that—a member of the flock. In
order to stay part of the flock, amember has an urge to be close to other members of the flock, thus working