Adaptive Learning of Human Driving Behavior for Nonlinear Vehicle Systems via Fuzzy Modelling

Artificial intelligence hopes machines perform tasks in a way naturally like a human being, a useful approach is behavior cloning (BC), which makes the machine learn human behavior. This paper addresses the online learning problem of human behavior based on a nonlinear human-in-the-loop (HiTL) vehicle system, where the human driver is modeled by a fuzzy controller. First, a Takagi-Sugeno (T-S) fuzzy model is constructed for the nonlinear vehicle control system. And a control model based on the parallel distributed compensation (PDC) technique is used to describe the human driver's behavior. A concurrent learning (CL)-based online adaptive law is then designed to learn the fuzzy feedback gain matrices of the fuzzy control model. Compaired with traditional approaches, the CL-based law removes the persistent excitation (PE) condition, mekes it more suitable for complex nonlinear systems. Finally, the simulation results are provided to demonstrate the effectiveness of the approach.