Abstract
Over the years, there has been a notable increase in the number of autonomous vehicles. One of the important tasks that self-driving cars need to accomplish is predicting steering angle. This project aims to improve the steering angle prediction of autonomous vehicles using lane detection techniques. Predicting the steering angle allows autonomous vehicles follow a particular lane. Lane detection extracts the critical features of the images used for steering angle prediction, i.e., lanes. Extracting lanes is important for improving the performance of steering angle prediction since the lane determines the angle the car needs to steer. Two approaches are investigated to detect the lanes in the images: one is an image-processing-based approach, and the other is Convolutional Neural Network (CNN)-based approach. In addition, two different CNN models (i.e., a CNN model based on NVIDIA’s model and a custom model) are used to predict the steering angle. The results show that lane detection using CNN significantly improves the performance of steering angle prediction models.