Experimental evaluation of adaptive model-based control for underwater vehicles in the presence of unmodeled actuator dynamics

Abstract

This paper reports a comparative experimental evaluation of proportional derivative and adaptive modelbased control for underwater vehicles. To the best of the authors' knowledge, this is the first such evaluation of modelbased adaptive tracking control for underwater vehicles during simultaneous dynamic motion in all 6 degrees-of-freedom. This experimental evaluation revealed the presence of unmodeled thruster dynamics arising during reversals of the vehicle's thrusters, and that the unmodeled thruster dynamics can destabilize parameter adaptation. The three major contributions of this paper are: an experimental analysis of how unmodeled thruster dynamics can destabilize parameter adaptation, a twostep adaptive model-based control algorithm which is robust to the thruster modeling errors present, and a comparative experimental evaluation of adaptive model-based control and proportional derivative control for fully-actuated underwater vehicles preforming simultaneous 6 degree-of-freedom trajectory tracking. I. INTRODUCTIONThis paper addresses the problem of adaptive modelbased trajectory tracking control of Underwater Vehicles (UV) for dynamic 6 degree-of-freedom (DOF) motion. The approach of adaptive model-based control (AMBC) refines estimates of plant parameters during the trajectory tracking control process. The adaptive controller estimates parameters for a rigid-body plant such as vehicle mass with added hydrodynamic mass, quadratic drag, gravitational force, and buoyancy torque parameters that arise in the dynamic models of rigid-body UV. We report a comparative experimental evaluation of AMBC and proportional derivative control (PDC) in full scale vehicle trials utilizing the Johns Hopkins University Hydrodynamic Test Facility and Johns Hopkins University Remotely Operated Vehicle (JHU ROV) (Fig. 1). This experimental evaluation shows that AMBC provides 30% better position tracking performance and 8% worse velocity tracking performance over PDC. To the best of the authors' knowledge, this is the first experimental comparison of AMBC and PDC during simultaneous motion in all DOF.

Justification for the Body Construction Selection of the Unmanned Uninhabited Underwater Apparatus

“…The system of six differential equations of motion can be used to calculate the parameters of the apparatus motion, as it is usually done when calculating the parameters of ships movement [2]. Due to the symmetry of the body, the equations of motion and the expressions for hydrodynamic characteristics used in them will be significantly simplified.…”

“…The calculation of the required thrust of the propellers can be performed by the formula [2] -propeller diameter.…”

“…For maneuvering in the vertical plane it is supposed to use the movement of the center of gravity of the apparatus by changing the position of the cargoes inside the apparatus. It is known that the cargo movement within the system [2], leads to the occurrence of the moment causing the trim or roll of the apparatus, depending on the direction of the cargo movement and the accepted system of designations. The mechanism of the system operation will be the same.…”

“…The shape of the contours, as a rule, is determined by the purpose of the apparatus. Most studies on UUV (Unmanned Underwater Vehicles) reveal practical operating experience or hydrodynamic research results [2][3][4].…”

A method of state estimation for underwater vehicle navigation around a cylindrical structure

“…Due to the difficulty of determining the exact dynamics of a complex-shaped underwater robot especially in the openwater environment with varying water current, the empirical plant dynamics (e.g., added mass, nonlinear Coriolis models, hydrodynamic damping forces, etc.) is usually employed as a proxy ( [10], [11]). Parameter identification can be done by computational fluid dynamics software ( [12], [13]) or practical experiments (e.g., [14]), but in this way only the rough values of plant parameters can be obtained.…”