Adaptive Look-Ahead Distance Based on an Intelligent Fuzzy Decision for an Autonomous Vehicle

Abstract

Autonomous vehicles use a set of perceptual and localization data proceeding from sensor measurements, in order to plan a certain trajectory based on decision-making, and finally to track the generated path. Trajectory following is performed by adjusting the steering angle generated by a lateral controller based on a geometric or non-geometric approach. The objective of the lateral controller is to minimize the lateral error between the vehicle and the path at a target point at a look-ahead distance from the vehicle. This paper investigates the look-ahead distance due to its high impact on performance alteration, and energy consumption. An intuitive analysis will be performed to study the effect of three varying parameters on the look-ahead distance, and the necessity to consider them. Then, a Fuzzy Logic approach will be established to adjust the look-ahead distance in accordance with three parameters: longitudinal velocity, road curvature, and original consideration of road adherence. Finally, a non-geometric model-based lateral controller will be developed based on the Super-Twisting Sliding Mode Control technique to control the steering angle via the Active Front Steering. The membership functions and the rules of the inputs and output of the Fuzzy Logic approach are implemented in a Matlab/Simulink environment and tested on a validated full non-linear vehicle model. Simulation results indicate the effectiveness of the fuzzy decision approach on the performance and energy consumption of the autonomous vehicle.