Distributed Control of Multiple Dynamic Systems with Resources Constraints and Its Application to Mobile Robot Formation Control

Description

Many applications, such as security and surveillance, environmental monitoring, and intelligent transportation, cannot be easily achieved by a single dynamic system platform if not impossible. A number of single dynamic systems, cooperating with each other through a communication network, are often needed for such applications. Examples of such systems, usually called multi-agent systems, include mobile robot groups, unmanned air vehicle teams, and wireless sensor networks. Each single dynamic system in the group/network is called a subsystem or an agent. In the past two decades, controlling each subsystem to support the cooperative and coordinated behavior of the entire system via distributed control strategies has been a focus in the robotics and control systems communities. The distributed control approach uses only local information. In recent years, controlling each subsystem with intermittent communication, instead of continuous communication, has become equally important. Researchers including the PI have been studying both problems intensively. However, most control strategies developed share a fundamental need to use either some global information about the entire system or continuous communication between neighboring agents, which inevitably leads to excessive consumption of the energy and communication resources. The main objective of this project is to develop a framework for handling the cooperative control problem of multi-agent systems with resources constraints via a fully distributed control strategy. Specifically, this project aims to develop novel adaptive distributed control laws, design aperiodic sampling and communication mechanisms, and conduct stability and consensus analysis of the closed-loop multi-agent systems with different agent dynamics, static or switching communication topologies, parameter uncertainties, and/or external disturbances. In this framework, with the novel adaptive distributed control strategies and well-designed aperiodic sampling and communication mechanisms, no global information will be needed, and information between neighboring agents will be exchanged only when certain conditions are satisfied. In such a way, it is expected that robustness, adaptively, and efficiency of the closed-loop multi-agent systems will be significantly enhanced. The study will rely on several tools from algebraic graph theory, matrix analysis, and control theory. The developed control strategies will be consolidated and validated via their application to mobile robot formation control. The results from this project will overcome many theoretical hurdles to meet the demanding challenges facing multi-agent systems in real applications. In particular, the results will improve not only the multi-agent systems’ resilience to communication loss and hardware failure, but also their utilization efficiency of energy and communication resources.