Metodologia de estimação paramétrica robusta aplicada a um veículo autônomo de superfície

Abstract

It is known that in running water activities or when exposed to winds and waves, the Autonomous Surface Vehicles (ASVs) are exposed to the most challenging scenarios. Under these conditions, knowledge of a robust mathematical model is a fundamental aspect for adjusting the control system. Therefore, it is a basic premise to achieve safety and performance in uncertain scenarios. In this context, this work presents a new identification methodology based on the use of multiple identification signals. In this perspective, the identification experiment is composed of optimized APRBS-type signals. These are designed to excite different samples around the initial estimation of the system and are optimized by an approach composed of the Particle Swarm Optimization (PSO) and the Interior-Point Method (IPM). Through this artifice, we seek to expand the excitation capacity of the experiment and, therefore, provide the estimation of more robust and reliable parametric models. To verify the effectiveness of the methodology, a catamaran-type ASV modeling problem was chosen. The results showed high similarity between the estimated models and the real vessel at all stages, including the various validation tests. Therefore, demonstrating the applicability of the proposed method and its ability to produce more robust and reliable parametric models.