摘要：Abstract: In this paper we analyze the application of a binary trust-consensus protocol in multi-agent systems with switching communication topology. Trust represents belief of one agent that the other one is capable of executing a specific task. The approach is motivated by an underwater robot system in which robots, called aMussels, are dispersed in the environment and measure environmental properties (for example sea-current) over a longer period of time, with sensors which are prone to failures. The goal is to apply the binary trust-consensus protocol to enable the system to autonomously detect failures of aMussels. Since direct communication between aMussels is energy consuming, a new type of moving robot, called aFish, is introduced that serves as a carrier of information. Since aFish have a limited communication range, communication topology is switching. Simulations are run first in Matlab for a homogeneous system, and then in a realistic simulation environment for the described scenario with heterogeneous agents. Results show that the protocol converges to the correct trust values and that the convergence time depends on the connectivity of the communication graph.
