Item

Abstract

Pure feedback and pure open-loop feedforward helicopter pi lot models are currently applied for predicting the performance of pilot-helicopter systems. We argue that feedback models are likely to underestimate performance in many realistic helicopter maneuvers, whereas inverse simulation models, which have an open-loop feedforward structure, are likely to overestimate performance as they neglect typical human-in-the-loop characteristics. True verification of feedback and feedforward elements in helicopter pilot control behavior was never performed, however. This paper proposes a pilot model containing a feedback and feedforward controller acting simultaneously and presents a method to identify the hypothesized feedforward action from human-in-the-loop data collected in a simulator experiment. The results of the human-in-the-loop experiment show that actual human performance is better than predicted by a pure feedback model and worse than predicted by an (inverse dynamics) feedforward model. The identification results suggest that the human pilot indeed utilizes feedforward strategies, but it was not possible to either confirm or refute the model by means of the collected data and the developed analysis method.