Robust model predictive control using optimized Laguerre functions: Application to autonomous articulated vehicles

Abstract

This thesis proposes innovative solutions for an advanced model predictive control (MPC) algorithm and an autonomous articulated vehicle system. The study has two main objectives. Firstly, it proposes an advanced model predictive control algorithm to ensure high performance and robustness with low computational complexity. Conventional MPC algorithms have limitations in terms of computational complexity and vulnerability to modeling errors. The proposed algorithm, named MPC using optimized Laguerre functions (OLMPC), significantly reduces computational complexity through input parameterization using Laguerre functions. Moreover, the Laguerre functions are optimized in real-time, guaranteeing high performance. The OLMPC approach is extended to various control problems, such as tracking control. A Robust OLMPC (ROLMPC) is proposed to enhance robustness against modeling errors. The robustness of ROLMPC is demonstrated by proving robust recursive feasibility and asymptotic stability under certain assumptions. Like this, this paper introduces a theoretically innovative and advanced MPC algorithm. Secondly, the paper proposes an obstacle avoidance system for autonomous articulated vehicles using ROLMPC. The system consists of trajectory planning and tracking control algorithms. The proposed trajectory planning algorithm optimizes path, velocity, and obstacle avoidance timing using single quadratic programming with novel weak duality functions. This approach significantly reduces the computational complexity compared to previous nonlinear programming approaches. The tracking control algorithm is designed using MPC, and a lumped lateral dynamics model of articulated vehicles is proposed as a predictive model. Unmeasurable states and parameters are estimated in real-time to improve the accuracy of the proposed model. An autonomous articulated vehicle system is constructed by integrating trajectory planning and tracking control algorithms. The proposed system ensures robustness by considering potential errors in the integrated system. The performance of the proposed system is evaluated through simulations, demonstrating its ability to effectively avoid obstacles in both normal and emergency driving situations, even in the presence of disturbances.