AdvGrasp: Adversarial Attacks on Robotic Grasping from a Physical Perspective

TL;DR

AdvGrasp introduces a physical perspective framework for adversarial attacks on robotic grasping, focusing on deforming objects to challenge lift capability and stability, thereby revealing system vulnerabilities.

Contribution

This work presents AdvGrasp, a novel method that systematically deforms objects to evaluate and improve the robustness of robotic grasping systems from a physical standpoint.

Findings

AdvGrasp effectively degrades grasp performance in diverse scenarios.

The method demonstrates robustness and practical applicability in real-world tests.

Abstract

Adversarial attacks on robotic grasping provide valuable insights into evaluating and improving the robustness of these systems. Unlike studies that focus solely on neural network predictions while overlooking the physical principles of grasping, this paper introduces AdvGrasp, a framework for adversarial attacks on robotic grasping from a physical perspective. Specifically, AdvGrasp targets two core aspects: lift capability, which evaluates the ability to lift objects against gravity, and grasp stability, which assesses resistance to external disturbances. By deforming the object's shape to increase gravitational torque and reduce stability margin in the wrench space, our method systematically degrades these two key grasping metrics, generating adversarial objects that compromise grasp performance. Extensive experiments across diverse scenarios validate the effectiveness of AdvGrasp, while real-world validations demonstrate its robustness and practical applicability