MACBeth: A Multi-Agent Constraint-Based Planner

Robert P. Goldman, Karen Zita Haigh, David J. Musliner, Michael Pelican

Honeywell Technology Center, Minneapolis, MN

2000 AAAI workshop "Constraints and AI Planning", pages 11-17

MACBeth is a multi-agent constraint-based tactical planning engine. It is designed for domains in which a time-pressured human user, such as a pilot or soldier, must quickly customize a mission for a team of autonomous agents. Through a playbook graphical user interface, the human user communicates instructions to the planner in the form of desired goals, tasks, partial plans or constraints. MACBeth evaluates these instructions for feasibility and correctness, and expands and refines them to produce fully executable plans. In tactical planning, the difficulty lies in how to apply familiar methods to novel situations, how to coordinate multiple agents so that the methods are carried off successfully, and how to manage limited resources. MACBeth directly addresses these issues, and has been used in two robotics domains: Unmanned Combat Aerial Vehicles (UCAV) sorties and Tactical Mobile Robotics (TMR) deployment. MACBeth provides the essential planning capabilities for these demanding domains.

  • Compressed Postscript (7 pages, 534K)
  • A copy of the entire proceedings, and other AAAI publications can be found at AAAI Press.