|
|||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||
ABSTRACT
This paper introduces Petri net (PN) based models of cooperative robotic tasks, namely those involving the coordination of two or more robots, thus requiring the exchange of synchronisation messages, either using explicit (e.g., wireless) or implicit (e.g., vision-based observation of teammates) communication. In the models, PN places represent primitive actions, subtasks and predicates set by sensor readings and communicated messages. Events are associated to PN transitions. The PN models can be used for task planning, plan execution and plan analysis. Different PN views enable the analysis of different properties. In this work we focus on plan analysis, namely on properties such as boundedness and liveness, corresponding to checking if resources usage is stable and plans have no deadlocks, as well as on stochastic performance, concerning the plan success probability. One novel feature of our work is that the analysis consists of composing several small action PN models with environment PN models, leading to a closed loop robot team/environment analysis methodology. Examples of application to simulated robotic soccer scenarios are presented. REFERENCES
Note: OCR errors may be found in this Reference List extracted from the full text article. ACM has opted to expose the complete List rather than only correct and linked references.
INDEX TERMS
Primary Classification:
Additional Classification:
General Terms:
Keywords:
|
|||||||||||||||||||||||||||||||||||||||||||