An architecture to integrate planning and execution in dynamic environments

In this paper we describe the architecture of a system for domainindependent planning in dynamic environments with incomplete information. The architecture permits many agents to work simultaneously on the same environment, which causes the planning to be a difficult task since the planner knowledge about the world becomes obsolete in a short time. Our approach is designed to overcome this problem through a highly coupled system composed of an incremental planning and an executor. Moreover, the system includes other important features in order to deal with real-world domains: sensing actions, a monitoring mechanism, planning in time-limited situations, use of numeric functions and a versatile graphical simulator.