Adversarial Robotics

Abstract: Robots act in adversarial environments. It is a fact. Unfortunately, little research has been done on strategic robotic behavior considering the existence of an adversary. This talk summarizes work that we have recently done in the new, emerging, area of Adversarial robotics: Accounting for adversarial presence in robotic tasks. This will be demonstrated by three different problems: multi-robot patrolling, robotic coverage and robot-team formation. We have shown that considering an adversary leads to a more general problem, where operating in neutral environments (as has been done so far) is actually an instance of this problem, that assumes a specific (usually simple, random) adversary.