Abstract
This paper proposes a closed-loop decentralised framework for swarm distribution guidance, which disperses homogeneous agents over bins to achieve a desired density distribution by using feedback gains from the current swarm status. The key difference from existing works is that the proposed framework utilises only local information, not global information, to generate the feedback gains for stochastic policies. Dependency on local information entails various advantages including reduced inter-agent communication, a shorter timescale for obtaining new information, asynchronous implementation, and deployability without a priori mission knowledge. Our theoretical analysis shows that, even utilising only local information, the proposed framework guarantees convergence of the agents to the desired status, while maintaining the advantages of existing closed-loop approaches. Also, the analysis explicitly provides the design requirements to achieve all the advantages of the proposed framework. We provide implementation examples and report the results of empirical tests. The test results confirm the effectiveness of the proposed framework and also validate the robustness enhancement in a scenario of partial disconnection of the communication network.
Highlights
We propose a closed-loop framework that relies on the Local Information
Consistency Assumption (LICA), i.e. only local information needs to be consistently known by the local agent groups
Exploiting LICA enables an asynchronous implementation of the framework and provides robustness against dynamical changes in bins as well as in agents
Summary
We propose a closed-loop framework that relies on the Local Information. The stability and performance of the proposed framework are extensively investigated via theoretical analysis and empirical tests. We prove that the agents asymptotically converge towards the desired swarm distribution even using local information-based feedback. Empirical tests demonstrate the performance of the proposed framework in three implementation examples: (1) travelling cost minimisation; (2) convergence rate maximisation under flux upper limits; and (3) quorum-based policies generation [similar to Halasz et al (2007), Hsieh et al (2008)]. We show an asynchronous version of the proposed framework and demonstrate that it is more robust against sporadic network disconnection of partial agents, compared with the recent work in Bandyopadhyay et al (2017)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.