Partially Ordered Planning with Extended Stubborn Set Pruning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated planning methods face challenges with exponential increases in computational complexity and search space due to the number of states and actions, leading to inefficiencies in determining optimal action sequences.
Innovation Solution
The method transforms planning problems into single goal planning problems, generates stubborn and extended stubborn sets, determines a pruned search space, and uses a planner algorithm to identify solutions within this space, effectively addressing the limitations of traditional top-quality planning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional automated planning methods are used to determine optimal action sequences, then complete search space exploration is performed, but computational complexity increases exponentially with the number of states and actions
Solution Approach 1:
The patent extracts and removes redundant states from the search space by identifying and eliminating states that cannot lead to optimal solutions. This is achieved through techniques such as removing dominated states (states that are strictly worse than other states) and pruning states that violate optimality conditions, thereby reducing computational complexity while preserving solution completeness.
Solution Approach 2:
The patent applies partial search by focusing computational resources only on necessary portions of the search space. Instead of exhaustively exploring all possible states, the method performs partial expansion of states based on heuristic guidance and optimality criteria, examining only those states that have potential to lead to optimal solutions.
2Productivity
If the search space is pruned to reduce computational complexity, then solution determination becomes faster, but risk of missing optimal solutions increases
Solution Approach 1:
The patent performs preliminary actions by pre-computing and storing dominance relationships between states, as well as pre-identifying pruning criteria before the actual search begins. This preparation work enables rapid pruning decisions during search without compromising optimality, as the necessary conditions for safe pruning are established in advance.
Solution Approach 2:
The patent incorporates feedback mechanisms where the search algorithm continuously monitors solution quality and search progress, using this information to dynamically adjust pruning decisions. The algorithm provides feedback to the pruning process to ensure that states critical for optimality are preserved while redundant states are eliminated.
Data Source
AI summary
Efficiently solving partially ordered top-quality planning includes receiving a first input associated with a planning problem. The planning problem is transformed into a single goal planning problem based on the reception of the first input. At least one stubborn set associated with the single goal planning problem is generated. Based on the at least one stubborn set, one or more extended stubborn sets associated with the single goal planning problem are determined. The one or more extended stubborn sets include at least one task action of a set of task actions associated with the single goal planning problem. A pruned search space associated with the single goal planning problem is determined based on the one or more extended stubborn sets. Based on the pruned search space, a set of solutions associated with the planning problem are determined and further rendered.