Optimizing Constrained Search Patterns for Remote Mine-Hunting Vehicles

When conducting remote mine-hunting operations with a sidescan-sonar-equipped vehicle, a lawn-mowing search pattern is standard if no prior information on potential target locations is available. Upon completion of this initial search, a list of contacts is obtained. The overall classification performance can be significantly improved by revisiting these contacts to collect additional looks. This paper provides, for the first time, a link between the recent literature which finds optimal secondary looks and optimal route planning software. Automated planning algorithms are needed to generate multiaspect routes to improve the performance of mine-hunting systems and increase the capability of navies to efficiently clear potential mine fields. This paper introduces two new numerical techniques designed to enable current remote mine-hunting systems to achieve secondary paths minimizing the total distance to be traveled and satisfying all motion and imaging constraints. The first "local" approach is based on a sequential algorithm dealing with more tractable subproblems, while the second is "global" and based on simulated annealing. These numerical techniques are applied to two test sites created for the Mongoose sea trial held at the 2007 Autonomous Underwater Vehicle (AUV) Fest, Panama City, FL. Highly satisfactory planning solutions are obtained.