Hidden Markov Model Based Control Augmentation Design for a Class of Human-in-the-Loop Systems.

Control augmentation can significantly boost the performance of systems with human-in-the-loop. However, the benefit of such designs has yet been fully realized because many parameters of human internal vehicle models are inaccurate. Here, a control augmentation framework is studied to assist human operators in controlling a system to precisely follow desired commands. There are two steps involved in this framework: (1) a Hidden Markov Model based estimator for unknown parameters in a human internal vehicle model; and (2) a regulator based on the identified human internal vehicle model to reduce tracking errors. A general form of the human internal vehicle model is applied to describe the operator’s understanding about the system dynamics. A recursive, closed-form solution is derived for a class of dynamical systems so that the computational cost can be significantly reduced. The algorithm is validated in a simulated, pilot-in-the-loop quadrotor scenario. [ABSTRACT FROM AUTHOR]