The computation of relaxed plans provides valuable informa-tion about the solution to a planning problem in polynomial time. Moreover, in some domains relaxed plans can be di-rectly taken as part of the solution saving expensive search episodes. Unfortunately, given that the construction of re-laxed plans ignores the delete effects of actions, relaxed plans may present flaws. In this paper, we propose a novel tech-nique for repairing flaws in relaxed plans, based on domain-specific rules learned from experience. The paper presents and evaluates three different relational learning approaches to automatically induce domain-specific rules from examples. The three learning approaches correspond to the planning systems ROLLER, CABALA and REPLICA that took part in the IPC-2008 learning track.

A Relational Instance-Based Policy can be defined as an action policy described following a relational instancebased learning approach. The policy is represented with a set of state-goal-action tuples in some form of predicate logic and a distance metric: whenever the planner is in a state trying to reach a goal, the next action to execute is computed as the action associated to the closest state-goal pair in that set. In this work, the representation language is relational, following the ideas of Relational Reinforcement Learning. The policy to transfer (the set of state-goal-action tuples) is generated with a planning system solving optimally simple source problems. The target problems are defined in the same planning domain, have different initial and goal states to the source problems, and could be much more complex. We show that the transferred policy can solve similar problems to the ones used to learn it, but also more complex problems. In fact, the policy learned outperforms the planning system used to generate the initial state-action pairs in two ways: it is faster and scales up better.

Current planners show impressive performance in many real world and artificial domains by using planning (either domain dependent or independent) heuristics. But, on one hand, domain dependent planners still outperform domain independent planners by re-defining domain theories, also including control knowledge. On the other hand, these domain dependent planners require a careful and manual refinement of domain theories to incorporate domain and control knowledge. Here, we present a tool that automatically generates domain and control knowledge as a middle ground solution to the definition of efficient quality-based planners.

AI planning solves the problem of generating a correct and efficient ordered set of instantiated activities, from a knowledge base of generic actions, which when executed will transform some initial state into some desirable end-state. There is a long tradition of work in AI for developing planners which make use of heuristics which are shown to improve their performance in many real world and artificial domains. The developers of planners have chosen between two extremes when defining those heuristics. The domain-independent planners use domain-independent heuristics, which exploit information only from the “syntactic” structure of the problem space and of the search tree. Therefore, they do not need any “semantic” information from a given domain in order to guide the search. From a Knowledge Engineering (KE) perspective, the planners that use this type of heuristics have the advantage that the users of this technology need only focus on defining the domain theory and not on defining how to make the planner efficient (how to obtain “good” solutions with the minimal computational resources). On the other hand, the domain-dependent planners require users to manually represent knowledge not only about the domain theory, but also about how to make the planner efficient.

The computation of relaxed plans provides valuable information about the solution to a planning problem in polynomial time. Moreover, in some domains relaxed plans can be directly taken as part of the solution saving expensive search episodes. Unfortunately, given that the construction of relaxed plans ignores the delete effects of actions, relaxed plans may present flaws. In this paper, we propose a novel technique for repairing flaws in relaxed plans, based on domainspecific rules learned from experience. The paper presents and evaluates three different relational learning approaches to automatically induce domain-specific rules from examples. The three learning approaches correspond to the planning systems ROLLER, CABALA and REPLICA that took part in the IPC-2008 learning track.