This paper challenges the prevailing pessimism about the scalability of partial order planning (POP) algorithms by presenting several novel heuristic control techniques that make them competitive with the state of the art plan synthesis algorithms. Our key insight is that the techniques responsible for the efficiency of the currently successful planners--viz., distance based heuristics, reachability analysis and disjunctive constraint handling--can also be adapted to dramatically improve the efficiency of the POP algorithm. We implement our ideas in a variant of UCPOP called REPOP # . Our empirical results show that in addition to dominating UCPOP, REPOP also convincingly outperforms Graphplan in several "parallel" domains. The plans generated by REPOP also tend to be better than those generated by Graphplan and state search planners in terms of execution flexibility. 1

In most real-world reasoning problems, planning and scheduling phases are loosely coupled. For example, in project planning, the user comes up with a task list and schedules it with a scheduling tool like Microsoft Project. One can view automated planning in a similar way in which there is an action selection phase where actions are selected and ordered to reach the desired goals, and a resource allocation phase where enough resources are assigned to ensure the successful execution of the chosen actions. On the other hand, most existing automated planners studied in Artificial Intelligence do not exploit this loose-coupling and perform both action selection and resource assignment employing the same algorithm. The current work shows that the above strategy severely curtails the scale-up potential of existing state of the art planners which can be overcome by leveraging the loose coupling. Specifically, a novel planning framework called RealPlan is developed in which resource allocatio...

We deal with the search process of the Graph-Plan algorithm in this paper. We concentrate on the problem of finding supports for a sub-goal which arises during the search. We model the problem of finding supports as a constraint satisfaction problem in which arc-consistency is maintained. Contrary to other works on the similar topic, we do not model the whole planning problem as a CSP but only a small sub-problem within the standard solving process. Our model is based on dual views of the problem which are connected by channeling constraints. We performed experiments with several variants of propagation in the constraint model through channeling constraints. Experiments confirmed that the dual view of the problem enhanced with maintaining of arc-consistency is a good choice.

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.

Knowledge about domain dynamics, describing how certain actions affect the world, is called an action model, and constitutes the essential requirement for planning and goal-oriented intelligent behaviour. Action learning, as the automatic construction of action models, has become a hot research topic in recent years, and various methods employing wide variety of AI tools have been developed. Diversity of these methods naturally renders each of them usable under different conditions and in various kinds of domains. After the extensive analysis of related work, we have declared our goal to introduce a collection of tractable and online methods for probabilistic action learning in noisy and partially observable domains supporting the induction of action’s preconditions and complex conditional effects. Subsequently, we have proposed the first solution candidate, which embeds our compact representation structure called effect formula (EF) and a polynomial algorithm 3SG (Simultaneous Specification, Simplification, and Generalization),