Learning-based Intelligent Attack against Formation Control with Obstacle-avoidance

Abstract

The security issue of mobile robots have attracted considerable attention in recent years. Most existing works focus on detection and countermeasures for some classic attacks from cyberspace. Nevertheless, those work are generally based on some prior assumptions for the attacker (e.g., the system dynamics is known, or internal access is compromised). A few work are delicated to physical attacks, however, there still lacks certain intelligence and advanced control design. In this paper, we propose a physical-based and intelligent attack framework against the obstacle-avoidance of mobile robots. The novelty of our work lies in the following: i) Without any prior information of the system dynamics, the attacker can learn the detection area and goal position of a mobile robot by trial and observation, and the obstacle-avoidance mechanism is learned by support vector regression (SVR) method; ii) Considering different attack requirements, different attack strategies are proposed to implement the attack efficiently; iii) The framework is suitable for holonomic and non-holonomic mobile robots, and the algorithm performance analysis about time complexity and optimality is provided. Furthermore, the condition is obtained to guarantee the success of the attack. Simulations illustrate the effectiveness of the proposed framework.The authors are with the

I Can Read Your Mind: Control Mechanism Secrecy of Networked Dynamical Systems under Inference Attacks

“…Once these elements of control mechanism are disclosed to the attacker, the robot system will face more severe security risks. For instance, the attacker can use the knowledge to launch oneshot strike against the critical robot precisely to disrupt the whole system [16], or choose to be a spy robot that sneaks into the system and stealthily misguides the system [17], [18].…”

Sneak Attack against Mobile Robotic Networks under Formation Control

“…These attacks are against a specific transducer by utilizing its sensing mechanism, and are hard to be generalized to other scenarios. [11] developed a trial-and-learning based…”

“…Lemma 1 shows that r i can be steered to any state by utilizing the characteristic of g. Given the input configuration of g, the avoidance behavior is unique. Therefore, as in [11], we make r a actively excite on r i and observe its external interaction response to learn g.…”

“…Several prior works have made some progress in this direction and support our demonstration. For example, it is possible to infer whether two robots are connected [10] or learn how a single robot avoids an obstacle [11]. However, considering what exact interaction rules can be learned and under what conditions to obtain them, how to further destroy the cooperation performance by leveraging the rules and where is the attack impact boundary, they are all still open issues and need further exploration.…”