Top-K AI Planning via Iterative Solution Forbidding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for top K planning are inefficient, requiring a long time to find the first solution and being restricted to searching in the state space using a K* search algorithm with admissible and consistent heuristic, which does not work well on many problems.
Innovation Solution
An iterative method that obtains and modifies planning problems to forbid found solutions, repeating the process until a desired number of solutions is found, using existing optimal planners and reformulating the planning task to preserve all solutions except for the given ones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If K* search algorithm with admissible and consistent heuristic is used for top K planning, then the search is restricted to state space, but the time required to find the first solution becomes excessively long
Solution Approach 1:
The patent segments the top-K planning process into multiple iterations, where each iteration finds one solution and then modifies the problem to forbid that solution. This divides the complex task of finding K solutions simultaneously into K sequential sub-tasks, significantly reducing the time to find the first solution while maintaining solution quality through systematic exploration of the state space in each iteration.
Solution Approach 2:
The patent applies preliminary action by modifying the planning problem beforehand in each iteration to forbid previously found solutions. This pre-modification ensures that subsequent searches automatically avoid known solutions, streamlining the search process and reducing redundant exploration, thereby decreasing the time required to discover new high-quality solutions.
2Productivity
If iterative method with problem modification is used to forbid found solutions, then solution generation becomes incremental, but the complexity of modifying the planning problem increases
Solution Approach 1:
The patent changes parameters of the planning problem by adding constraints that forbid specific solutions. Instead of fundamentally altering the problem structure, it modifies numerical or logical parameters (such as adding forbidden state constraints) that guide the search away from known solutions while preserving the core planning task, thus balancing productivity improvement with manageable complexity.
3Ease of manufacture
If existing optimal planners are used for each iteration, then implementation becomes simpler, but the ability to handle diverse planning problems may be limited
Solution Approach 1:
The patent achieves universality by designing an iterative framework that can wrap around any existing optimal planner. The framework itself becomes a multi-functional system that adapts to different planning problems by configuring the underlying planner appropriately, thus maintaining ease of implementation through reuse of existing tools while gaining versatility through the adaptive iterative structure.
Data Source
AI summary
A method for improving performance of at least one hardware processor solving a top-k planning problem includes obtaining, in a memory coupled to the at least one processor, a specification of the planning problem in a planning language; obtaining, in a first iteration carried out by the at least one processor, at least one solution to the planning problem; and modifying the planning problem, in the first iteration carried out by the at least one processor, to forbid the at least one solution. The method further includes repeating, by the at least one processor, the obtaining of the at least one solution and the modifying to forbid the at least one solution, for a plurality of additional iterations, after the first iteration, until a desired number, k, of solutions to the planning problem are found or until no further solutions exist, whichever comes first.