Majid Abbasi

Yazd University, Iran

Abstract Title: The Inverse IRL Control Problem (Inverse Reinforcement Learning), a model-free data-driven inverse IRL algorithm without knowing fully system model for aircraft control application

Biography:

Majid Abbasi is a doctoral student at Yazd University and has completed his master's degree from Sistan and Baluchistan University. He is a member of the Faculty of Mathematics, the Azad University of Yazd.

Research Interest:

In this paper is introduced the fundamental framework of inverse optimal control in continuous-time (CT) optimal control problems with infinite horizon costs, which offers an alternative perspective to traditional optimal control methods. Optimal control aims to design a controller that minimizes a given performance index. This often requires solving the steady-state Hamilton– Jacobi–Bellman (HJB) equation or the algebraic Riccati equation for linear systems, which can be intractable in some cases. Inverse optimal control has emerged as a promising approach that circumvents the complexity associated with solving the HJB equation. In real-world applications, the system models may not be known. To solve the Inverse IRL Control Problem (Inverse Reinforcement Learning), a model-free data-driven inverse IRL algorithm is presented without knowing fully system model knowledge based on the model-based algorithm. Simulation examples (f16 aircraft) show the effectiveness of this algorithm.