AUV Optimal Path for Leak Detection

Olivier Marceau,Jean-Michel Vanpeperstraete

DOI: https://doi.org/10.48550/arXiv.1801.02377

2018-01-08

Systems and Control

Abstract:This paper studies an optimal autonomous underwater vehicule (AUV) path planning method for both reducing average delay before pollutants detection in underwater mining, oil or gas fields and reducing AUV occupancy time. The proposed technique, based on the bayesian search theory framework and multi-objective optimization, extracts optimal boustrophedon paths for leak detection in complex environment. We describe a multi-objective nonlinear mixed integer optimization model for both reducing global nondetection probability and path duration. We then propose a hierarchical algorithm combining two functions. The main function is a multi-objective cross entropy which places the tracklines. The second function sets the optimal speeds on each trackline by means of an interior point method. Numerical simulations show that the proposed framework is a very promising approach because the optimal paths cross spill of highly probable leaks before less probable ones. We show that our optimized paths outperform boustrophedon paths of same duration with uniform speed and spacing of trackline. Thanks to Pareto efficiency approach, our tool propose optimal trajectories for numerous AUV autonomies. Hence it can be used for both real time path planning and design purpose.