Original Theoretical Contributions to Stochastic Optimal Control for Autonomous Underwater Vehicle Path Planning in Turbulent Niger Delta Estuarine Environments

Autonomous underwater vehicle operations in the complex estuarine environment of the Niger Delta are subjected to stochastic hydrodynamic disturbances from tidal currents, surge from offshore wave propagation, and fluvial flows from the distributary network that substantially degrade path-following performance of deterministic control approaches designed for calmer open-ocean deployments. This dissertation presents original theoretical contributions to stochastic optimal control for AUV path planning under the specific uncertainty characteristics of Niger Delta estuarine flow fields. A stochastic ocean current model for Niger Delta estuaries is developed and calibrated from acoustic Doppler current profiler data collected at nine sites across the Forcados and Ramos distributary channels over 14 months. A risk-sensitive stochastic optimal control formulation is proposed that incorporates the AUV dynamic model, the stochastic current model, and a risk sensitivity parameter allowing explicit trade-off between expected energy expenditure and variance of mission completion time. The resulting Hamilton-Jacobi-Bellman equations are solved numerically using a sparse grid approximation method developed as part of this dissertation to address the curse of dimensionality at the six-dimensional AUV state space. Comparative sea trial experiments using an ECA A9-E AUV demonstrate that the proposed stochastic optimal controller reduces path-following cross-track error by 38 percent and energy consumption by 19 percent relative to a robust deterministic controller under typical estuarine current conditions. A probabilistic collision risk metric for AUV navigation near submerged pipeline infrastructure is derived and integrated into the planning framework. Keywords: stochastic optimal control, AUV path planning, Niger Delta, estuarine hydrodynamics, underwater robotics