A systematic method for dynamic modeling and identification of a small-sized autonomous surface vehicle using simulated annealing techniques

Abstract

This paper presents and validates a method for the dynamic modeling and identification of an Autonomous Surface Vehicle (ASV) taking into account the performance of the sensors usually installed in relatively low-cost surface vehicles. For the estimate of the parameters of the adopted model we propose an onboard sensor-based, off-line identification procedure based on Simulated Annealing. The method is systematic and was successfully applied to produce a nonlinear model of a robotic catamaran belonging to the HydroNet ASV class [1]. The used sensors consisted of a compass measuring the heading angle and a paddle wheel speed sensor to measure the robot surge speed: GPS data were not used during the identification to loose the need of a relatively expensive GPS receiver to produce an accurate model. Extended experiments at sea showed our approach is effective in producing a dynamic model of the ASV. It uses data produced by inexpensive sensors in sea trials without needing costly facilities such as tow-tanks or planar motion mechanisms. The resulting model proved sufficiently accurate to be a valid support to simulation and control law design.