Multi-agent plan reconfiguration under local LTL specifications
2015 (English)In: The international journal of robotics research, ISSN 0278-3649, E-ISSN 1741-3176, Vol. 34, no 2, 218-235 p.Article in journal (Refereed) Published
We propose a cooperative motion and task planning scheme for multi-agent systems where the agents have independently assigned local tasks, specified as linear temporal logic formulas. These tasks contain hard and soft sub-specifications. A least-violating initial plan is synthesized first for the potentially infeasible task and the partially-known workspace. This discrete plan is then implemented by the potential-field-based navigation controllers. While the system runs, each agent updates its knowledge about the workspace via its sensing capability and shares this knowledge with its neighbouring agents. Based on the knowledge update, each agent verifies and revises its motion plan in real time. It is ensured that the hard specification is always fulfilled for safety and the satisfaction for the soft specification is improved gradually. The design is distributed as only local interactions are assumed. The overall framework is demonstrated by a case study and an experiment.
Place, publisher, year, edition, pages
2015. Vol. 34, no 2, 218-235 p.
Autonomous agents, distributed robot systems, networked system, cognitive robotics, agent-based systems
Computer Vision and Robotics (Autonomous Systems)
IdentifiersURN: urn:nbn:se:kth:diva-161116DOI: 10.1177/0278364914546174ISI: 000349144800005ScopusID: 2-s2.0-84921953956OAI: oai:DiVA.org:kth-161116DiVA: diva2:797129
FunderSwedish Research CouncilEU, FP7, Seventh Framework Programme, RECONFIG: FP7-ICT-2011-9-600825
QC 201503232015-03-232015-03-092015-03-23Bibliographically approved