In this paper, the genetic algorithm (GA), which is a simulation of Darwinian evolution and an efficient way for large-scale optimization subject to non-linear constraints, is applied to find economical and safe routes for a swarm of AUVs to revisit an area with waypoints and obstacles which are known a priori. The algorithm can be divided into three phases: (1) waypoint assignment: allocating the waypoints to individual AUVs; (2) route optimization: minimizing the total journey of the vehicles and (3) route validation: checking whether there exist stationary and/or moving collisions. A case study for three AUVs to survey a given area is also presented to verify the algorithm.