Constrained Fractional-Order Model Predictive Control for Robust Path Following of FWID-AGVs With Asymptotic Prescribed Performance

• Published in: IEEE transactions on vehicular technology Vol. 74; no. 2; pp. 2692 - 2705
• Main Authors: Zhao, Jing, Li, Renbin, Zheng, Xinyang, Li, Wenfeng, Hu, Chuan, Liang, Zhongchao, Wong, Pak Kin
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.02.2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accuracy; Asymptotic methods; Asymptotic properties; Autonomous ground vehicle; Closed loops; Computational modeling; Constraints; Controllers; Cost function; Disturbances; fractional-order; Land vehicles; Liapunov functions; model predictive control; Neural networks; Optimal control; path following; Predictive control; Predictive models; prescribed performance control; Recursive functions; Robust control; Robustness; Steady-state; Tires; Trajectory planning; Unmanned ground vehicles; Vehicle dynamics; Wheels
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2024.3476921

This paper proposes a prescribed performance function (PPF)-based constrained fractional-order model predictive controller (CFMPC) for the robust path following of the four-wheel-independent-drive autonomous ground vehicle (FWID-AGV) subject to the model mismatch and external disturbances. First, a PPF-based model predictive controller (MPC) is developed to guarantee the transient-state performance by dealing with the model mismatch. Second, a contraction constraint of the MPC is constructed with an adaptive neural network backstepping method, aiming to ensure the optimal control inputs fall into the region of attraction. Third, a fractional-order cost function of the MPC is discretized to reject the external disturbances and improve the robustness of the controller. Besides, a sufficient condition is given to prove the recursive feasibility of the PPF-based CFMPC, and a Lyapunov function candidate is derived for the asymptotic stability of the closed-loop path following system. The effectiveness and practicability of the proposed method are validated by hardware-in-the-loop tests, respectively.