Preferences Optimization track
This track concerns planning with soft goals, which are desired goals that a valid plan does not have to necessarily achieve. Not achieving a goal implies a certain penalization added to the cost of the plan. Cost of the plan can be just the addition of the penalizations or some combination of the total actions cost and penalizations (as in last IPC net-benefit track). Planners competing in this track are expected to find optimal plans in terms of the objective function. Planners compare on the number of problems they manage to solve optimally within a reasonable amount of time.
- Core features: STRIPS, action costs, goal utilities, metric quantities.
- Optional features: ADL, derived predicates.
- Objective function: Minimize total cost.
Core features are those that will appear in all of the domains. So, in order for a planner to have chances to win this track of the competition, all the core features must be supported. The performance of all planners will be assessed in all domains they can run.
Optional features concern alternative but equivalent versions of each domain. These features will be used only in cases where they facilitate domain modelling. When several alternative versions of a domain are supported by a planner, it is scored on the version on which it performs best.