A school of fish exhibits remarkable emergent behaviors: it maneuvers swiftly,
it evades predators and it forages successfully. Biologists are developing models
that can reproduce school behaviors with simple traffic rules for individual
fish. Our aim is similar but our problem is different: to design coordinated
dynamics for a network of autonomous vehicles. The vehicle network will serve
as a reconfigurable sensor array capable of efficient search and discovery.
Because our prescribed traffic rules will be implemented on real vehicles, we
guarantee stability, scalability and robustness of the schooling dynamics. Our techniques make use of artificial potentials, virtual bodies, symmetry and reduction. In a current application, we are designing coordinated dynamics for a fleet of underwater gliders that will be deployed
in Monterey Bay as part of an autonomous ocean sampling network.