This paper reviews the connections between Graphplan's planning-graph and the dynamic constraint satisfaction problem and motivates the need for adapting CSP search techniques to the Graphplan algorithm. It then describes how explanation based learning, dependency directed backtracking, dynamic variable ordering, forward checking, sticky values and random-restart search strategies can be adapted to Graphplan. Empirical results are provided to demonstrate that these augmentations improve Graphplan's performance significantly (up to 1000x speedups)on several benchmark problems. Special attention is paid to the explanation-based learning and dependency directed backtracking techniques as they are empirically found to be most useful in improving the performance of Graphplan. 1. Introduction Graphplan (Blum & Furst, 1997) is currently one of the more efficient algorithms for solving classical planning problems. Four of the five competing systems in the recent AIPS-98 planning comp...

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...

Recent work has demonstrated that treating resource reasoning separately from causal reasoning can lead to improved planning performance and rational resource management where increase in resources does not degrade planning performance. However, the resources were scheduled procedurally and limited to cases that could be solved backtrackfree. Terming the decoupled framework as RealPlan, in this work, I extend it with a general approach to convert the resource allocation problem as a declaratively specified dynamic constraint satisfaction problem (DCSP), compile it into CSP and solve it with a CSP solver. By doing so, the resource scheduling problem can be handled in its full complexity and can provide a computational characterization of the different scheduling classes. The CSP formulation also facilitates planner-scheduler interaction by helping the scheduler interpret the resource allocation policies proposed by the planner in terms of constraints on values of schedul...

Despite their near dominance, heuristic state search planners still lag behind disjunctive planners in the generation of parallel plans in classical planning. The reason is that directly searching for parallel solutions in state space planners would require the planners to branch on all possible subsets of parallel actions, thus increasing the branching factor exponentially. We present a variant of our state search planner AltAlt called AltAlt-p which generates parallel plans by using greedy online parallelization of partial plans. The greedy approach is significantly informed by the use of novel distance heuristics that AltAlt-p derives from a graphplan-style planning graph for the problem. While this approach is not guaranteed to provide optimal parallel plans, empirical results show that AltAlt-p is capable of generating good quality parallel plans without losing also the quality in terms of the total number of actions in the solution at a fraction of the time cost incurred by the disjunctive planners. Keywords:Domain-independent planning, scalability in planning, heuristic state search planning, parallel plans, and partial planning.

Tackling real-world problems often requires to take various types of constraints into account. Such constraint types range from simple numerical comparators to complex resources. This article describes how planning techniques can be integrated with general constraintsolving frameworks, like SAT, IP and CP. In many cases, the complete planning problem can be cast in these frameworks. 1

Abstract. Reformulation of a problem is intended to make the problem more amenable to efficient solution. This is equally true in the special case of reformulating a planning problem. This paper considers various ways in which reformulation can be exploited in planning. 1

Planning consists of 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. In most realworld problems, these two phases are loosely coupled. On the other hand, most existing planners 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 planners, including such recent ones as Graphplan and Blackbox. In response, a novel planning framework was developed in which resource allocation is de-coupled from planning and is handled in a separate "scheduling" phase. Implementing this framework raises several interesting issues regarding the role of resources in planning, the interactions between the planning and scheduling phases and the choices in selecting the meth...

Searching for parallel solutions in state space planners is a challenging problem, because it would require the planners to branch on all possible subsets of parallel actions, exponentially increasing their branching factor. We introduce a variant of our heuristic state search planner AltAlt, which generates parallel plans by using greedy online parallelization of partial plans. Empirical results show that our online approach outperforms postprocessing (offline) techniques in terms of the quality of the solutions returned.

The main concern in automated planning is to construct a sequence of actions that achieves an objective given an initial situation of the world. Planning is hard; even the most restrictive case of automated planning, called classical planning, is PSPACE-complete in general. Factors affecting planning complexity are large search spaces, problem decomposition and complex action and goal interactions. One of the most straightforward algorithms employed to solve classical planning problems is state-space search. In this algorithm, each state is represented through a node in a graph, and each arc in the graph corresponds to a state transition carried out by the execution of an action from the planning domain. A plan on this representation corresponds to a path in the graph that links the initial state of the problem to the goal state. The crux of controlling the search involves providing a heuristic function that can estimate the relative goodness of the states. However, extracting heuristic functions that are informative, as well as cost effective, remains a challenging problem. Things get complicated by the fact that subgoals comprising a state could have complex interactions. The specific contributions of this work are:

Planning as satisfiability is a principal approach to planning with many eminent advantages. The existing planning as satisfiability techniques usually use encodings compiled from STRIPS. We introduce a novel SAT encoding scheme (SASE) based on the SAS+ formalism. The new scheme exploits the structural information in SAS+, resulting in an encoding that is both more compact and efficient for planning. We prove the correctness of the new encoding by establishing an isomorphism between the solution plans of SASE and that of STRIPS based encodings. We further analyze the transition variables newly introduced in SASE to explain why it accommodates modern SAT solving algorithms and improves performance. We give empirical statistical results to support our analysis. We also develop a number of techniques to further reduce the encoding size of SASE, and conduct experimental studies to show the strength of each individual technique. Finally, we report extensive experimental results to demonstrate significant improvements of SASE over the state-of-the-art STRIPS based encoding schemes in terms of both time and memory efficiency. 1.