Exploiting Structure in AI Planning

Automated planning, a core area of Artificial Intelligence, deals with choosing and organising actions to achieve an objective at least cost. Applications include space exploration, crisis management, military operations, and manufacturing. The greatest obstacle to realising the potential of planning technology is the inefficiency caused by the lack of systematic exploitation of the structure of the planning problem at hand. The project will identify powerful notions of problem structure, and build new structure-aware heuristics and planners upon them. This will enable automated planners to rival the efficiency of domain-specific action selection tools, and reduce the cost of developing decision-making software in the application areas.