Baike She1, Siddhartha Mehta2, Emily Doucette3, J. Willard Curtis4, Zhen Kan5
11:00 - 11:20 | Wed 10 Jul | Franklin 4 | WeA04.4
A leader group selection approach that jointly considers network controllability and control energy is investigated in this work. Specifically, a dynamic multi-agent system with signed acyclic topology is considered, where signed edges capture cooperative and competitive interactions among agents. To effectively and efficiently control the multi-agent system, leader group selection in this work focuses on energy-related controllability, which jointly considers two primary objectives: 1) network controllability, i.e., identification of a subset of agents as leaders that can drive the entire network to a desired state even in the presence of antagonistic interactions, and 2) energy efficiency, which takes into account the control cost incurred by the selected leaders in steering the network to the desired state. To achieve these objectives, graph-inspired characterizations of energy-related controllability are developed based on the interaction between the network topology and the agent dynamics. The developed topological characterizations are exploited to derive heuristic leader selection algorithms on signed acyclic graphs. Illustrative examples are provided to demonstrate the effectiveness of the developed leader group selection methods.