Across Viewpoints and Time: Trajectory-Generalizable Physical Attacks on World Models and Video Understanding Models
cs.CV, cs.AI
Submitted: 2026-10-11
Updated: 2026-10-11
Terminology
Sources
- V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning
- Adversarial Patch
- Tex3D: Objects as Attack Surfaces via Adversarial 3D Textures for Vision-Language-Action Models
- UniTexture: Cross-Task Universal Adversarial Textures for Vision-Language-Action Models
- When World Models Dream Wrong: Physical-Conditioned Adversarial Attacks against World Models
- WMAttack: Automated Attack Search for Adversarial Evaluation of World-Model Agents
- Targeting World Models to Compromise Robot Learning Pipelines
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models