Comparison of Mathematical Models for the HUGIN 4500 AUV Based on Experimental Data

Abstract

This paper reports the development and comparison of several mathematical models for describing the response of the HUGIN 4500 AUV as a function of measured actuator inputs. As a crucial step in fitting the derived models to experimental field data, a combination of dedicated experiments and least-squares estimation is applied in order to obtain an estimate of the sea current, and properly compensate for it during the identification stage. The established models are extensively cross-validated with uncorrelated data, which is a standard step in system identification to prevent over-fitting observed data. The vehicle response predicted by the proposed models shows good agreement with real measurements. The possible use of mathematical vehicle models includes control and navigation system design, applications to fault detection and integrity, and vehicle design evaluation. The content of this paper is motivated from the perspective of developing second generation underwater navigation systems, tightly integrating model-based velocity estimates and knowledge of commanded and measured actuator signals.