15:50 - 16:10 | Mon 19 Aug | Lau, 5-206 | MoC5.2
Inspired by cooperative foraging behavior in goose flocks, this paper proposes a distributed least-informed (DLI) method to achieve energy-efficient flocking control by selecting the least required informed agents. To avoid the fragmentation issue when applying the DLI method, three new concepts about the informed path, cycle, and tree are proposed. Based on the properties of the proposed new concepts, a theorem of fragmentation prevention is developed for multi-agent systems with a portion of informed agents. Utilizing the developed theorem of fragmentation prevention, the initialization and updating principles of the selection of the least informed agents in the DLI method are described. Simulation results demonstrate that by using the DLI method, a multi-agent system of 10 agents can successfully achieve energy-efficient flocking control without fragmentation. Namely, a collective motion around a (virtual) leader with velocity consensus (flock formation) is achieved by a total 15.9% energy reduction, compared with the completed informed method.