Bottom-up motion and task coordination for loosely-coupled multi-agent systems with dependent local tasks
2015 (English)In: IEEE International Conference on Automation Science and Engineering, IEEE , 2015, 348-355 p.Conference paper (Refereed)Text
We propose a bottom-up motion and task coordination scheme for loosely-coupled multi-agent systems under dependent local tasks. Instead of defining a global task for the whole team, each agent is assigned locally a task as syntactically co-safe linear temporal logic formulas that specify both motion and action requirements. Inter-agent dependency is introduced by collaborative actions of which the execution requires multiple agents' collaboration. The proposed solution contains an offline initial plan synthesis, an on-line request-reply messages exchange and a real-time plan adaptation algorithm. It is distributed in that any decision is made locally based on local computation and local communication within neighboring agents. It is scalable and resilient to agent failures as the dependency is formed and removed dynamically based on the plan execution status and agent capabilities, instead of pre-assigned agent identities. The overall scheme is demonstrated by a simulated scenario.
Place, publisher, year, edition, pages
IEEE , 2015. 348-355 p.
Automation, Computation theory, Distributed computer systems, Linear temporal logic, Local communications, Local computation, Loosely coupled, Multiple agents, Plan adaptation, Plan synthesis, Task coordination, Multi agent systems
IdentifiersURN: urn:nbn:se:kth:diva-181138DOI: 10.1109/CoASE.2015.7294103ISI: 000380453000056ScopusID: 2-s2.0-84952776941ISBN: 9781467381833OAI: oai:DiVA.org:kth-181138DiVA: diva2:901649
11th IEEE International Conference on Automation Science and Engineering, CASE 2015, 24 August 2015 through 28 August 2015
QC 201602082016-02-082016-01-292016-08-23Bibliographically approved