Study on intelligent vehicle steering control algorithm using SVM

Abstract

The traditional steering control algorithms require accurately dynamic and mechanic model which are very difficult. Aiming at the intelligent vehicle model, a vision-based support vector machine (SVM) steering control algorithm for autonomous navigation of vehicle is brought forward in this paper. With the input of white-line position values of a road map, the algorithm outputs the steering angle of the front wheels. 190 groups of image data is trained by SVM, which contain 20 groups of linear state models as well as 170 groups of turning state model. The test indicates that the algorithm can learn the control tactic of the operator very well, and has better stability and robustness. © 2013 Springer Science+Business Media.