Fuzzy logic control with adaptive methods for vehicle lateral guidance

In this dissertation lateral motion control of a vehicle is investigated in the framework of an automated highway system (AHS). The objective of lateral motion control for lane following is to achieve accurate tracking of a reference lane while maintaining an acceptable level of passenger comfort in the presence of disturbances and over a range of operating conditions. The proposed controllers, based on fuzzy logic control (FLC) theory, are developed based on implicit models of the vehicle dynamics as opposed to explicit mathematical models required by many conventional control design techniques.; The structure of the proposed FLC is modularized into feedback, preview, and gain scheduling rule bases. In the initial control design the parameters of the FLC are tuned manually using information from characteristics of human driving operation and an existing controller. Three feedback FLCs with different feedback variables are designed. A fuzzy preview rule base is developed to utilize preview information regarding the upcoming radii of curvature. Also, a gain scheduling rule base is designed to choose the appropriate controller based on the velocity of the vehicle and the traction conditions between the road and the vehicle tires.; In order to achieve an acceptable level of performance, methods are considered to investigate the automatic tuning of the FLC parameters. The first method uses genetic algorithms (GA) and the second approach uses a novel idea to adjust the FLC parameters on-line to follow a reference closed-loop model, modeled by a fuzzy system. In the ladder method, using Lyapunov theory, a supervisory control term is introduced to ensure that the closed-loop system under fuzzy logic control will maintain stability in the sense that the states of the vehicle are bounded by a specified maximum value.; Experimental test results of the manually tuned FLCs are shown, and a comparison is made to similar tests conducted using a frequency shaped linear quadratic (FSLQ) controller as well as a proportional, integral, plus derivative (PID) controller.