Controlling Basin Breakout for Robots Operating in Uncertain Flow Environments

Abstract

We present the development and experimental validation of an autonomous surface/underwater vehicle (ASV/AUV) control strategy that leverages the environmental dynamics and noise to efficiently navigate in a stochastic fluidic environment. In this work, we assume the workspace is composed of the union of a collection of convex regions, each bounded by Lagrangian coherent structures (LCS). LCS are dynamical features in the flow field that function like invariant manifolds in general non-autonomous dynamical systems and they denote regions in the flow field where more escape events occur. We show through theory and simulation that a vehicle’s likelihood of transition between adjacent LCS-bounded regions can be manipulated by the proposed control strategy, resulting in effective navigation strategies from one region to another. In addition, we show how optimal escape trajectories with respect to the transition probability between adjacent LCS-bounded regions can be determined. These trajectories correspond to energy-efficient trajectories since they leverage the inherent dynamics of the surrounding flow field. We experimentally show that the proposed control strategy exhibits a predictable exponential scaling of escape times and is effective even in situations where the structure of the flow is not fully known, there exist significant stochastic fluctuations, and control effort is costly.