Building a Plan with Real-Time Execution Guarantees

The degree to which a planning system succeeds depends on its ability to meet critical deadlines as well as the correctness and completeness of its models which describe events and actions that change the world state. It is often unrealistic to expect either unlimited execution time or perfect models, so a planner must be able to make appropriate time vs. quality tradeoffs, then detect and respond to states it had not originally planned to handle. In this paper, we consider these issues in the context of the Cooperative Intelligent Real-time Control Architecture (CIRCA), which combines a planner with a separate real-time system so that plans are built, scheduled, and then executed with real-time guarantees. Specifically, we discuss our recent addition of a probabilistic model to help the planner prioritize states for expansion, and present important classes of “unplanned-for” states that we detect and handle in CIRCA. Finally, we describe our current work to improve CIRCA’s planner by estimating planning time constraints in advance and incorporating a more intelligent utility function to prioritize states.