A Formal Metareasoning Model of Concurrent Planning and Execution

• DOI: 10.1609/aaai.v37i10.26464
• Publication Date: 2023-06-27T17:56:09Z
• Authors: Amihay Elboher; Ava Bensoussan; Erez Karpas; Wheeler Ruml; Shahaf S. Shperberg; Eyal Shimony
• Keywords: FOS: Computer and information sciences; Artificial Intelligence (cs.AI); Computer Science - Artificial Intelligence; 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology

Agents that plan and act in the real world must deal with fact time passes as they are planning. When timing is tight, there may be insufficient to complete search for a before it act. By commencing execution concludes, one gains by making planning concurrent. However, this incurs risk of incorrect action choices, especially if actions irreversible. This tradeoff between opportunity problem addressed paper. Our main contribution formally define setting an abstract metareasoning problem. We find intractable. we identify special cases solvable polynomial time, develop greedy solution algorithms, and, through tests on instances derived from problems, several methods achieve promising practical performance. work lays foundation principled time-aware executive concurrently plans executes.