Datensatz

Zusammenfassung

This work considers the problem of designing optimal multi-agent trajectories to patrol an environment. In both civil and military applications it is of increasing importance
to instruct a team of autonomous agents to accomplish
repetitive tasks, such as the monitoring of strategic regions
or the detection of life threatening situations. As performance
criterion for optimal patrolling we minimize the worst-case time
gap between any two visits of an environment location. We
characterize the computational complexity of the trajectory
design (patrolling) problem with respect to the environment
topology and to the number of robots to be employed in
the surveillance task. Even though the patrolling problem is
generally NP-hard, we identify particular cases that are solvable
efficiently, and we describe optimal patrolling trajectories.
Finally, we present a heuristic with performance guarantee, and
an 8-approximation algorithm to solve the NP-hard patrolling
problem.