Future of AD Security: Addressing Limitations and Ethical Concerns in Typographic Attack Research
1 Oct 2025
This paper summarizes a comprehensive framework for typographic attacks, proving their effectiveness and transferability against Vision-LLMs like LLaVA
Empirical Study: Evaluating Typographic Attack Effectiveness Against Vision-LLMs in AD Systems
1 Oct 2025
This article presents an empirical study on the effectiveness and transferability of typographic attacks against major Vision-LLMs using AD-specific datasets.
Foreground vs. Background: Analyzing Typographic Attack Placement in Autonomous Driving Systems
1 Oct 2025
This article explores the physical realization of typographic attacks, categorizing their deployment into background and foreground elements
Exploiting Vision-LLM Vulnerability: Enhancing Typographic Attacks with Instructional Directives
30 Sept 2025
This article proposes a linguistic augmentation scheme for typographic attacks using explicit instructional directives.
Methodology for Adversarial Attack Generation: Using Directives to Mislead Vision-LLMs
30 Sept 2025
This article details the multi-step typographic attack pipeline, including Attack Auto-Generation and Attack Augmentation.
The Dual-Edged Sword of Vision-LLMs in AD: Reasoning Capabilities vs. Attack Vulnerabilities
30 Sept 2025
This article analyzes the critical safety trade-off of integrating Vision-LLMs into autonomous driving (AD) systems.
Autoregressive Vision-LLMs: A Simplified Mathematical Formulation
30 Sept 2025
Explaining the role of logits and the softmax function in converting the output vector into a final probability distribution for the next token.
The Vulnerability of Autonomous Driving to Typographic Attacks: Transferability and Realizability
30 Sept 2025
This article reviews and compares two major types of adversarial attacks against neural networks: gradient-based methods (like PGD) and typographic attacks.
The Integration of Vision-LLMs into AD Systems: Capabilities and Challenges
27 Sept 2025
This article reviews the development and application of Vision-Large-Language-Models, focusing on their integration into autonomous driving systems.