Observability-based optimization for flow sensing and control of an underwater vehicle in a uniform flowfield

This paper describes how an underwater vehicle can control its motion by sensing the surrounding flowfield and using the sensor measurements in a dynamic feedback controller. Limitations in existing sensing modalities for flowfield estimation are mitigated by using a fish-inspired distributed sensor array and a nonlinear observer. Estimation performance is further increased by optimizing sensor placement on the vehicle body. We optimize sensor placement along a streamlined body using measures of flowfield observability, namely the empirical observability gramian. Velocity potentials model the flow around the vehicle and a recursive Bayesian filter estimates the flow from noisy velocity measurements. To orient the body into the oncoming flow (a fish-inspired behavior known as rheotaxis) we implement a dynamic, linear controller that uses the estimated angle of attack. Numerical simulations illustrate the theoretical results.