Field Results of the Control, Navigation, and Mapping Systems of a Hovering AUV

Abstract

This paper describes the evolution of the control, navigation, and mapping capabilities of a hovering autonomous underwater vehicle (AUV) designed to explore flooded cenotes in Mexico as part of the DEPTHX project. The vehicle is equipped with a suite of navigation sensors that allow it to localize itself and and create maps of the complex 3D environments in which it operates. It is passively stabilized in the roll and pitch directions and equipped with six thrusters that allow it to directly actuate each of the remaining degrees of freedom (x, y, z, and heading). We describe the control system of the vehicle, which provides open-water and near-wall manuevering capabilities. Near-wall behaviors are used in support of science operations such as visual wall survey and precise placement of a core sampling tool. We also present the vehicle's navigation system, which combines dead-reckoning, sonar-based localization, and simultaneous localization and mapping (SLAM). We demonstrate the performance of these systems using experimental results from multiple test environments, including a test tank, a flooded limestone quarry, and the La Pilita cenote. Trajectory plots comparing desired and actual motions will be used to demonstrate the various vehicle controllers. Data from repeated localization experiments will be used to provide statistically meaningful measures of the accuracy of the dead-reckoning, localization, and SLAM filters.