作者: Edmund H. Durfee , Marcus J. Huber
DOI:
关键词: Joint (building) 、 Application domain 、 Computer science 、 Joint commitment 、 Commit 、 Imperfect 、 Work (electrical) 、 Risk analysis (engineering) 、 Abandonment (legal) 、 Simulation
摘要: It is important, in situations where teams of agents can commit to joint, long-term objectives, that the be able identify when team objective no longer or futile. Prior work has typically assumed should believe joint commitments hold unless explicitly told otherwise by other agents. In this paper, we argue assumption expects too much from communication channels and initiate abandonment commitment. A more general, robust method for monitoring a commitment make each agent responsible acquiring evidence supports refutes contention sustained While include explicit messages others, it also observations made interpretations those observations. We thus build formal notions collective goals, conventions maintenance beliefs about commitments, extend allow consideration imperfect channels. Our approach been implemented tested application domain cooperative robotic reconnaissance sometimes hostile environment.