Abstract & Overview
This thesis presents the design and developments of robotic and virtual SWARM testbeds within SWARM Laboratory at the Australian National University. Some experimental works and theoretical outcomes are also included as parts of thesis research. These two testbeds aim to implement the active researches within System and Control Group in ANU on multi-agent systems, especially on formation control, onto physical robots and improve the algorithms practically by making comparisons of the experimental results from robotic and virtual testbeds. Due to the lack of sensing and communicating capabilities of the physical robots, an alternative approach, which deploys visual aid, networked communication and remote control, is proposed to provide localization capabilities for agents to achieve the distributed control system scheme of robotic testbed. While, the virtual testbed are capable to conduct self-localization and inter-agent communication, hence, a different distributed system approach is proposed. A series of various drive modes are developed to simplify the process of implementing Matlab simulations and to improve algorithms' performance. Then a novel method is presented for future use of these two testbed, which enables quick implementations on testbeds with a few modifications on Matlab simulation codes. At last, theoretical research on the development of a cooperative shadowing system is presented, and a speed only approach within this framework is proposed and investigated on the virtual testbed system.