R&D of the maneuver control techniques for Autonomous Surface Vehicle using "Dual-extended Kalman filte" for the actual harbor area surveillance

In order to achieve the high-precision underwater exploration using Autonomous Underwater Vehicle (AUV), it is necessary to develop an intelligent Autonomous Surface Vehicle (ASV) which has a highly accurate underwater acoustic positioning system merged with precise GPS. Such ASV equips a localization method of AUV with a USBL positioning system, and collaboration method to control AUV with an acoustic communication system. Moreover, a highly accurate measurement in very-shallow-water, such as dam-lakes and a restricted water zones are difficult to observe by the ship-mounted measurement system manipulated by human. Therefore, it is necessary to develop an ASV that can accomplish the underwater measurements accurately and automatically to the programmed traverse line under the difficult environment such as tidal currents and winds. In such R&Ds, it is important to develop the automatic motion control for ASV which detects the external tidal flow automatically and handles precise ASV control to run on the programmed line. However, the conventional auto-pilot control methods that manipulate only the rudder angle and yawing angle of the course of the survey-way-points, have been difficult to cruise on the measurement line accurately because they don't have a feedback control system between the XY-plane kinematic model and yawing-dynamics model. In this paper, our R&D group proposes about a sophisticated ASV maneuver motion control technique in line with the programmed survey line for the collaborative underwater automatic measurement using ASVs and AUVs. The proposed technique uses two extended Kalman filters (EKFs) individually from motion equations of the plane-dynamics and ship-handling dynamics. The main factor which make difficult to control the rudder angle is the dynamic transverse flow. Our proposed technique estimates these values using EKF model of the plane motion including these flow environments. ASV calculates the drift distance from the survey-line based on planemotion EKF prediction, and also calculates the rudder controlled variables. After the rudder control, these values are estimated and predicted by the yawing-motion EKF and these predictions of heading are used to the plane-motion EKF. We checked our proposed control algorithm by using a simulation, and confirmed that this algorithm can cruise nearby the programmed line. And we developed the prototype ASV in this year, and confirmed that autonomous homing operation is available to the acoustic position data of underwater test towing body from the sea operation test.