This presentation introduces an optimization framework for 3D trajectory planning of autonomous underwater vehicles in uncertain currents. First, we describe incorporating currents from ocean general circulation models (OGCM) into a nonlinear programming 3D trajectory planning formulation. The OGCM current data provides 3D current vectors and enables the construction of an indicator function to avoid obstacles. Secondly, we propose a risk-aware ensemble-based stochastic optimization problem. We pose this problem as a stochastic programming formulation resembling a two-stage decision framework, where the path (sequence of waypoints without timing information) is a first-stage variable common to all ensemble members. The second-stage variables (also called recourse variables) defined for each ensemble member are the time steps, velocity vectors, and acceleration vectors. The objective function is a convex combination of the expectation operator and the conditional value at risk of travel time or energy consumption. The case studies with deterministic currents will demonstrate the optimization framework capabilities to find solutions that take advantage of the current and depth to minimize travel times and energy consumption. The case studies with uncertain currents compare solutions obtained with average currents with stochastic optimization solutions with different risk-aware levels.
Ricardo M. Lima is a Research Scientist in Professor Omar Knio’s research group in CEMSE. He is also the co-founder of the KAUST/Saudi startup Decision Science Technologies. Ricardo joined KAUST in 2014. He received the Ph.D. degree in 2006 in Chemical Engineering from the Faculty of Engineering, University of Porto, Portugal. In 2006, he became a Post-doc fellow in the Department of Chemical Engineering at the Carnegie Mellon University, Pittsburgh, PA, USA. From 2008 to 2011, he was an invited researcher in PPG Industries, USA. He was a Marie Curie Fellow in the National Laboratory of Energy and Geology (LNEG) in Lisbon, Portugal from 2011 to 2014. His research interests include modeling and optimization of complex problems related to chemical, processing industries, and energy systems. Target applications include integration, planning and scheduling of renewable energy systems, process synthesis, planning and scheduling of chemical engineering systems, and trajectory planning of autonomous underwater vehicles. In terms of methodologies, Ricardo focus on the development of mathematical programming methodologies, namely combinatorial optimization models, continuous optimization models, deterministic global optimization solution approaches, stochastic programming models, and decomposition algorithms to solve large-scale problems.