A closer look at “Robot Safety Systems Can Be Fooled by Manipulated Inputs, VicOne Researchers Warn”, exactly as the source publishes it. Even when a robot's safety functions are working as designed, manipulated sensor inputs could lead to unsafe behavior, according to researchers at VicOne LAB R7. In tests, a robot dog changed its movement after reading text on a poster as an instruction, and a simulated hospital service robot altere In tests, a robot dog changed its movement after reading text on a poster as an instruction, and a simulated hospital service robot altered its behavior in response to crafted audio inaudible to humans.
These demonstrations show how visual or audio inputs can be exploited to override a robot's intended task without changing its programming. The concern extends to independent protective systems that rely on sensor data to trigger safety stops. For the complete material, the listing points back to the source inside the network.