Driving through the Lens: Improving Generalization of Learning-based Steering using Simulated Adversarial Examples

Yu Shen, Laura Yu Zheng, Manli Shu, Weizi Li, Tom Goldstein, Ming Lin

Keywords: model robustness, data augmentation, adversarial training, image quality, autonomous driving, benchmark

Abstract: To ensure the wide adoption and safety of autonomous driving, the 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 of the vehicle. In this work, we address this critical issue by analyzing the sensitivity of the learning algorithm with respect to 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 self-driving cars.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

One-sentence Summary: Analysis the relationship between image quality factors and "learn to steer" performance, and propose a method to improve the robustness on different scenarios, together with a benchmark.

Reviewed Version (pdf): https://openreview.net/references/pdf?id=3oWLdBHK7