Collaborative control of multiple AUVs for improving the estimation of flow field dependent variables

Abstract

This paper presents a control framework for creating a 3D map of flow field dependent physical ocean variables by controlling multiple AUVs to maximize information gain. In this framework a nonlinear Kalman Filter is used to update the noisy measurements from multiple sensors at uncertain positions. First, the area of interest is discretized into a 3D grid. Each grid point has an associated estimate or measurement of the physical variable as well as its uncertainty. Multiple AUVs make measurements to update the value and uncertainty of the grid point at their current location. When this information is communicated between AUVs an updated 3D map is then propagated forward through time. To determine how to control the AUVs, a Nonlinear Model Predictive Controller (NMPC) is developed to generate paths for the AUVs which will minimize the overall uncertainty of the estimates in the 3D map of physical ocean variables. Simulations are shown with multiple AUVs to illustrate the utility and application of this approach. Various fluid dynamic environments, e.g. vortex flow, are initialized, and the AUVs are controlled to optimally measure a temperature distribution. The results show this method improves the estimation of the ocean variables as well as decreases mission time when compared to naive search methods.