Rationale-Supported Mixed-Initiative Case-Based Planning

Mixed-initiative planning envisions a framework in which automated and human planners interact to jointly construct plans that satisfy specific objectives. In this paper, we report on our work engineering a robust mixed-initiative planning system. Human planners rely strongly on past planning experience to generate new plans. ForMAT is a case-basedsystem that supports human planning through the accumulation of user-built plans, query-driven browsing of past plans, and several plan functionality analysis primitives. Prodigy/Analogy is an automated AI planner that combines generative and case-based planning. Stored plans are annotated with plan rationale and reuse involves adaptation driven bythis rationale. Our system, MI-CBP, integrates ForMAT and Prodigy/Analogy into a real-time message-passing mixed-initiative planning system. The main technical approach consists of allowing the user to specify and link objectives that enable the system to capture and reuse plan rationale. We prese...

• Type: Conference
• Year: 1997
• Where: AAAI
• Authors: Manuela M. Veloso, Alice M. Mulvehill, Michael T. Cox
• Keywords: AAAI 1997, Human Planners, Intelligent Agents, Mixed-initiative Planning, Mixed-initiative Planning System