Searching Realistic-Looking Adversarial Objects For Autonomous Driving Systems

TL;DR

This paper introduces a modified gradient-based texture optimization method with a 'Judge' agent to generate realistic-looking adversarial objects for autonomous driving systems, enhancing real-world applicability.

Contribution

It proposes a novel approach combining a 'Judge' agent with texture optimization to produce realistic adversarial objects, advancing prior simulation-to-real transfer research.

Findings

Strategies 2 and 3 are more promising for reliable realism assessment.

Strategies 1 and 4 are less reliable for judging realism.

The method improves the practical relevance of adversarial object generation.

Abstract

Numerous studies on adversarial attacks targeting self-driving policies fail to incorporate realistic-looking adversarial objects, limiting real-world applicability. Building upon prior research that facilitated the transition of adversarial objects from simulations to practical applications, this paper discusses a modified gradient-based texture optimization method to discover realistic-looking adversarial objects. While retaining the core architecture and techniques of the prior research, the proposed addition involves an entity termed the 'Judge'. This agent assesses the texture of a rendered object, assigning a probability score reflecting its realism. This score is integrated into the loss function to encourage the NeRF object renderer to concurrently learn realistic and adversarial textures. The paper analyzes four strategies for developing a robust 'Judge': 1) Leveraging cutting-edge vision-language models. 2) Fine-tuning open-sourced vision-language models. 3) Pretraining neurosymbolic systems. 4) Utilizing traditional image processing techniques. Our findings indicate that strategies 1) and 4) yield less reliable outcomes, pointing towards strategies 2) or 3) as more promising directions for future research.