GhostTac: Contactless Attack on Tactile Sensors via Electromagnetic Interference

Kun Wang and other authors propose GhostTac, a method for contactlessly manipulating robot tactile sensors through electromagnetic interference (EMI). The authors claim this is the first contactless attack of its kind: it exploits the nonlinear rectification and limited-bandwidth amplification effects of EMI to convert constructed EMI signals into a sustained DC bias, bypassing onboard filtering and causing stable measurement deviations. The authors validated it on 15 tactile sensors of different types across 10 sensor modules and 2 dexterous hands, claiming that all tested devices exhibited consistent attack effects; they also demonstrate practical impact through three case studies—tactile grasping, slip detection, and material classification—for example, potentially causing home robots to apply excessive force. The work has been accepted by ACM CCS 2026.

AI 根据报道生成 · 6 days ago更新