Secure Distributed Consensus Control for Multi-Robot Systems
Primary Investigator:
Byung-Cheol Min
Sangjun Lee and Byung-Cheol Min
Abstract
In this paper, we present distributed switching consensus control policies to achieve cyberattack-resilient multi-robot consensus. We present residual-based attack detection schemes against two major types of cyberattacks: deception attack and denial of service (DoS) attack, and distributed switching consensus protocols for countermeasures. For the deception attack, we utilize the coordination-free consensus protocols to adaptively tune the weights on signal strength measurements based on the attack profile. This allows multi-robot systems to achieve global consensus by assigning more weight to the robots in normal operation than the compromised robots. For the DoS attack, we utilize a leader-follower strategy that quickly reconfigures the communication topology to reassign the compromised robots as followers. This strategy ensures that there is at least another neighbor in normal operation within sensor range, allowing each robot to switch its local control law between predefined consensus protocols in a fully distributed fashion. The proposed consensus control is tested using the Robotarium multi-robot platform. Experimental results demonstrate that the proposed control enables the multi-robot team to achieve consensus in the presence of cyberattacks.