Direct Measurement Based H∞ Controller Synthesis for an Autonomous Surface Vehicle

An autonomous surface vehicle, based on a Prindle-19 catamaran and substituting a self-trimming vertical wing for the sail, was developed to demonstrate precision guidance and control. This vehicle, the Atlantis, was demonstrated to track straight line segments to better than 0.3 meters (1 − σ) when already trimmed for sail along the segment, using LQG control based on an identified plant using the Observer Kalman Identification (OKID) methods. In this work, a way-point guidance system is tested experimentally in addition to a novel H∞ subspace direct controller that is designed based on measured time series data for both the inputs and the actuator outputs. In previous simulations the Model Free Subspace H∞ controller has demonstrated similar performance levels to LQG methods while using the same identification data, but without requiring a model structure. Results from experimental trials have proven less successful, motivating an analysis of the controller’s implementation and possible problems.