Improving Robustness of Learning-based Autonomous Steering Using Adversarial Images

TL;DR

This paper presents a framework and algorithm to enhance the robustness of learning-based autonomous steering systems against varying image quality, improving performance by up to 48% and outperforming existing methods.

Contribution

The authors introduce a novel robustness analysis framework and an algorithm that significantly improves autonomous steering performance under diverse image conditions.

Findings

Performance improved by up to 48%

Outperforms data augmentation and adversarial training

Enhances robustness and generalization of neural networks

Abstract

For safety of autonomous driving, vehicles need to be able to drive under various lighting, weather, and visibility conditions in different environments. These external and environmental factors, along with internal factors associated with sensors, can pose significant challenges to perceptual data processing, hence affecting the decision-making and control of the vehicle. In this work, we address this critical issue by introducing a framework for analyzing robustness of the learning algorithm w.r.t varying quality in the image input for autonomous driving. Using the results of sensitivity analysis, we further propose an algorithm to improve the overall performance of the task of "learning to steer". The results show that our approach is able to enhance the learning outcomes up to 48%. A comparative study drawn between our approach and other related techniques, such as data augmentation and adversarial training, confirms the effectiveness of our algorithm as a way to improve the robustness and generalization of neural network training for autonomous driving.