Abstract

Navigating Autonomous Ground Vehicles (AGVs) demands handling complex traffic conditions with safety, comfort, and precision across diverse operating conditions. An essential constituent of AGV navigation is the development of effective path-following strategies to track a desired trajectory while dealing with dynamic variables such as changing vehicle speeds, external disturbances, and actuator limitations. Such systems must combine robust control methods to handle uncertainties and nonlinearities inherent in vehicle dynamics and provide reliable performance in complex driving conditions. This research focuses on developing a lateral control framework for AGVs through a novel Linear Parameter Varying (LPV) modeling approach. By treating the vehicle's longitudinal velocity as a varying parameter, an LPV representation of the road-vehicle system is formulated. To reduce the conservatism inherent in traditional LPV models, a polytopic LPV framework that employs a varying parameter and utilizes a first-order Taylor approximation is introduced. This approach reduces the number of vertices required in the polytopic vehicle model, thus improving computational efficiency and performance of the path tracking control design. This polytopic LPV model, employed for an H_2 LPV control design, has the dual objectives of disturbance attenuation and guaranteeing passenger safety and comfort. The control design process utilizes Lyapunov theory to guarantee the stability and performance of the closed-loop system. Both Quadratic Lyapunov Functions (QLF) and Poly-Quadratic Lyapunov Functions (PQLF) are explored, with PQLF suggesting a substantial reduction in the conservatism typically associated with LPV systems. The controller design is recast as a convex optimization problem under Linear Matrix Inequality (LMI) constraints. This method avoids the drawbacks of heuristic optimization methods while ensuring a systematic and robust solution. Additionally, a Linear Quadratic Regulator (LQR) control scheme is developed as a benchmark that is further validated through numerical simulations. The performance of the proposed controllers and LPV models is extensively assessed through comparative simulations with their nonlinear counterparts, which emphasizes their effectiveness in dealing with the dynamic challenges of AGV lateral control.