Distributed decision making for task switching via a consensus-like algorithm

We propose a consensus-like algorithm for switching from a first to a second task in multi-agent systems. The task-switch is supposed to occur synchronously for all agents after each single agent has finished the first task, because the two tasks may influence each other. As every agent has only local information about its own task-state, the proposed algorithm is used to determine the state of the whole network. With this additional information each agent can decide when to safely switch the task. In this paper, the first task is a distributed exploration control law, while the second task is a distributed coverage control law where the goal is to maximize the probability of detecting certain events. By exploring the mission space before switching to the coverage task, the overall probability of detecting those events is increased. For the exploration task a r-limited Voronoi cell based algorithm is used and for the coverage task a gradient based control law.