Generalized Optimal Control for Autonomous Mine Countermeasures Missions

Abstract

This article presents a computational framework for planning mine countermeasures (MCM) search missions by autonomous vehicles. It employs generalized optimal control (GenOC), a model-based trajectory optimization approach, to maximize the expected search performance of vehicle–sensor pairs in different minehunting scenarios. We describe each element of the proposed framework and adapt it to solve real-world MCM motion planning problems. A key contribution of this article develops sensor models that are more tunable than conventional ones based on lateral range curves. The proposed models incorporate engineering parameters and 3-D geometry to compute mine detection probability as a function of sonar design and search vehicle trajectories. Specific examples for various forward-looking and sidescan sonar systems deployed by unmanned vehicles are included. Objective computations utilize these sonar detection models during optimization to minimize the risk that candidate search trajectories fail to detect mines in an area of interest. Simulation results highlight the flexibility of our proposed GenOC framework and confirm that optimal trajectories outperform conventional search patterns under time or resource constraints. We conclude by identifying some of the practical considerations of this approach, and suggest ways that numerical analysis of GenOC solutions can be used for MCM mission planning and decision aid development.