Hijacking Vision-and-Language Navigation Agents with Adversarial Environmental Attacks

TL;DR

This paper presents a whitebox adversarial attack on vision-and-language navigation agents, showing how environmental modifications can hijack agent behavior, causing them to ignore instructions and follow attacker-defined paths.

Contribution

It introduces a novel 3D adversarial attack method that manipulates environment appearance to deceive pretrained VLN agents, highlighting vulnerabilities in embodied AI systems.

Findings

Attacks cause agents to ignore instructions and follow unintended paths.

Environmental modifications can induce early-termination or diversion behaviors.

Attacks significantly impair the agents' ability to follow user instructions.

Abstract

Assistive embodied agents that can be instructed in natural language to perform tasks in open-world environments have the potential to significantly impact labor tasks like manufacturing or in-home care -- benefiting the lives of those who come to depend on them. In this work, we consider how this benefit might be hijacked by local modifications in the appearance of the agent's operating environment. Specifically, we take the popular Vision-and-Language Navigation (VLN) task as a representative setting and develop a whitebox adversarial attack that optimizes a 3D attack object's appearance to induce desired behaviors in pretrained VLN agents that observe it in the environment. We demonstrate that the proposed attack can cause VLN agents to ignore their instructions and execute alternative actions after encountering the attack object -- even for instructions and agent paths not considered when optimizing the attack. For these novel settings, we find our attacks can induce early-termination behaviors or divert an agent along an attacker-defined multi-step trajectory. Under both conditions, environmental attacks significantly reduce agent capabilities to successfully follow user instructions.