Environmental Engineering Reference
In-Depth Information
gliders used for oceanographic research. The intent was to develop low-cost
planetary exploration systems that could run autonomously for years in harsh
environments, such as in the sulfuric acid atmosphere of Venus or on Titan
(the largest of Saturn's moons).
A second NASA swarm-related project titled “Extremely Large Swarm
Array of Picosats for Microwave/RF Earth Sensing, Radiometry, and Map-
ping” [
73
] was also funded by NIAC. The proposed telescope would be
used to do such things as characterize soil moisture content, atmospheric
water content, snow accumulation levels, flooding, emergency management
after hurricanes, weather and climate prediction, geological feature identifi-
cation, and others. To accomplish this would require an antenna size on the
order of 100 km at a GEO orbit. To implement such an large antenna, a
highly sparse spacefed array antenna architecture was proposed that would
consist of 300,000 picosats, each being a self-contained onechip spacecraft
weighing 20 g.
10.3 Other Applications of Swarms
The behavior of swarms of bees has also been studied as part of the BioTrack-
ing project at Georgia Tech [
9
]. To expedite the understanding of the behavior
of bees, the project videotaped the behavior of bees over a period of time, us-
ing a computer vision system to analyze data on sequential movements that
bees use to encode the location of supplies of food, etc. It is anticipated that
such models of bee behavior can be used to improve the organization of co-
operating teams of simple robots capable of complex operations. A key point
is that the robots need not have a priori knowledge of the environment, and
that direct communication between robots in the teams is not necessary.
Eberhart and Kennedy have developed an optimization technique based
on particle swarms [
78
] that produces fast optimizations for a wide number
of areas including UAV route planning, movement of containers on container
ships, and detecting drowsiness of drivers. Research at Penn State University
has focused on the use of particle swarms for the development of quantita-
tive structure activity relationships (QSAR) models used in the area of drug
design [
23
]. The research created models using artificial neural networks and
k-nearest neighbor and kernel regression. Binary and niching particle swarms
were used to solve feature-selection and feature-weighting problems.
Particle swarms have influenced the field of computer animation also.
Rather than scripting the path of each individual bird in a flock, the Boids
project [
115
] elaborates a particle swarm of simulated birds. The aggregate
motion of the simulated flock is much like that in nature. The result is from the
dense interaction of the relatively simple behaviors of each of the (simulated)
birds, where each bird chooses its own path.
Much success has been reported from the use of ant colony optimization
(ACO), a technique that studies the social behaviors of colonies of ants and
