Robust Control and Estimation Methods for Longitudinal and Lateral Motions of Autonomous Vehicle

Abstract

In this dissertation, we develop and analysis the robust control design methods for longitudinal and lateral motions of the autonomous vehicle. The estimator and the controller that are robust against changes in internal and external variables of the vehicle are proposed for longitudinal and lateral motions of autonomous vehicle. In addition,the kinematic and dynamic models are considered for the lateral motion of the autonomous vehicle. The robust localization and tracking schemes are also proposed and validated via computational simulations and experiments. First, the robust cooperative adaptive cruise control is proposed, which guarantees the string stability with heterogeneous vehicles. The augmented disturbance observer is designed for the uncertainty of the heterogeneous vehicles and the performance of the proposed method is validated via experiments. Second, we propose the dynamic model-based approaches for the lateral vehicle motion control and estimation. To overcome the unsynchronized update interval between sensors for measuring the driving environment and status of the vehicle, we decompose the dynamic model into two submodels in terms of the relations of each sensor, and we proposed the multi-rate state estimator. In addition, the estimator for lateral slip angle, which can not be directly measured by a sensor due to the physical limitations of the sensor, is designed. The robust backstepping control method is proposed using an augmented disturbance observer in order to solve the problem of computation time caused by model complexity and an external disturbance. Third, we develop the robust estimator and controller with the kinematic lateral motion model. The kinematic model-based robust lane-keeping system and lane restoration methods are first presented. The robust integrated lateral control method with empirical kinematic model is proposed for longitudinal and lateral motions to ensure the stability of entire closed-loop. Finally, a robust localization and tracking problem is considered. A linear parameter varying approach-based estimator is designed to solve the computational complexity. Also, a robust model predictive control design method satisfying the clothoid constraint is designed for robust waypoints tracking of an autonomous vehicle. The robust lane-keeping control scheme is presented to allow sensor failure for the autonomous vehicle. The results of the proposed methods in this dissertation provide guidelines for designing the estimator and controller of autonomous vehicle. The proposed methods are robust to disturbance, uncertain environment and varying parameters of a vehicle, and can overcome a practical problems of autonomous vehicles.