Conflict detection and resolution in collaborative planning

Abstract

In multi-agent collaborative planning, since each agent is autonomous and heterogeneous, it is inevitable that conflicts arise among the agents during the planning process. A collaborative agent, however, must be capable of detecting and resolving these conflicts. This paper describes a computational model that captures the collaborative planning process in a Propose-Evaluate-Modify cycle of actions. Our model is capable of evaluating a given proposal to detect potential conflicts regarding both proposed actions and proposed beliefs, and of initiating collaborative negotiation subdialogues to resolve the detected conflicts. In situations where multiple conflicts arise, our model identifies the focus of the modification process and selects appropriate evidence to justify the necessity for such modification. Finally, our model handles the negotiation of proposed domain actions, proposed problem-solving actions, and proposed beliefs in a unified manner.