Combined Lateral and Longitudinal Control with Variable Reference Path for Automated Driving

Abstract

This paper presents combined lateral and longitudinal trajectory control with variable reference path for autonomous vehicle. A longitudinal velocity controller is designed to determine whether the current velocity is appropriate for the real-time or planned path. A steering control algorithm has been constructed using Ackermann steering geometry based path tracking controller known as pure pursuit method. The pure pursuit algorithm calculates the steering angle using the characteristics of the Ackerman steering geometry of the vehicle. This algorithm is not suitable at high speed but robust against disturbance and is a good method to apply in normal autonomous driving situations. The risk of the current speed is analyzed through the predicted lateral acceleration. However, the predicted lateral acceleration will vary depending on the vehicle. Therefore, the accuracy of the predicted lateral acceleration is improved using an adaptive neural network algorithm. The proposed algorithm has been verified by computer simulation with two consecutive curvature roads in city driving. The verification results were quite satisfactory.