Ultra-local model predictive control: A model-free approach and its application on automated vehicle trajectory tracking

Zejiang Wang,Junmin Wang

DOI: https://doi.org/10.1016/j.conengprac.2020.104482

IF: 4.057

2020-08-01

Control Engineering Practice

Abstract:Model predictive control (MPC) has been extensively utilized in the automotive applications, such as autonomous vehicle path planning and control, hybrid-vehicle energy management, and advanced driver-assistance system design. As a typical model-based control law, MPC relies on a system model to predict the state evolution of the manipulated plant within the prediction horizon. However, a representative yet concise mathematical description of the controlled plant may not always be available in practice. Therefore, model-free strategies, e.g., identification for control and direct data-driven control, have been incorporated into the predictive control framework. Nonetheless, existing model-free predictive controllers usually require reliable datasets and employ complex nonconvex optimizations to identify the underlying system model. Furthermore, their control performances are fundamentally limited by the quality of the training data. Inspired by the model-free control, this paper proposes the ultra-local model predictive control (ULMPC), which is a novel and straightforward model-free predictive control technique with no need for the computationally-extensive model learning process. The proposed ULMPC is implemented for automated vehicle trajectory following. Carsim-Simulink joint simulations and indoor experimental field tests with a scaled car demonstrate its effectiveness.

automation & control systems,engineering, electrical & electronic