Genetic planning of control actions for a high manoeuvrability AUV

For certain operations the control of AUV motions is subject to difficulties, due to motion coupling. In this paper a high manoeuvrability AUV is considered, with only thrusters for motion control. A Genetic planning of the control actions has been devised, considering a multiobjective optimisation problem, which includes energy and time saving, and smooth operation with realistic control actions. Since the inertia of the submergible filters out control brisk changes, it turns out that there are many almost equivalent control strategies obtaining similar trajectories. This is why the characteristics of the control actions have been included as part of the multiobjective optimisation. Several diving scenarios have been specified, including obstacles. The paper begins with a description of the AUV and the control problem, then it focus on the multiobjective optimisation problem and the genetic planning to solve the problem, then it considers several scenarios and present optimal results; finally, the paper comes to conclusions and future research, which will be experimental.