Human Behavior Learning for A Class of Norm-Bounded Uncertain Linear HiTL Systems via Adaptive Inverse Optimal Guaranteed Cost Control

This paper presents an adaptive inverse optimal guaranteed cost control (IOGCC) approach for norm-bounded uncertain linear human-in-the-loop (HiTL) systems. The method addresses the challenge of learning human behavior modeled using an optimal guaranteed cost control (GCC) law. The quadratic cost weight matrix is initially unknown. Our approach consists of two steps. First, an adaptive law estimates the control gain matrix from system state data online, and a leakage term is incorporated to reduce the effect of model uncertainty. Second, using the learned control gain matrix, we solve a linear matrix inequality (LMI) optimization problem to identify the human cost function weight matrix. The effectiveness of the method is validated in a lane-keeping simulation.