Methods and systems for diverse instance generation in artificial intelligence planning
View Patent ↗Embodiments for creating planning tasks are provided. A plurality of atoms are generated. The plurality of atoms are partitioned into a plurality of variables. A casual graph is generated based on the plurality of variables. A layered graph including interchanging variable value layers and action layers is created based on the casual graph. A planning task is generated based on the layered graph.
1. A method for creating a planning task, by a processor, comprising:
generating a plurality of atoms;
partitioning the plurality of atoms into a plurality of variables;
generating a casual graph based on the plurality of variables;
creating a layered graph including interchanging variable value layers and action layers based on the casual graph, wherein the creating of the layered graph is ceased when the last of the variable value layers includes all of the plurality of atoms and the casual graph is fully covered; and
generating a planning task based on the layered graph according to a goal created for the planning task, wherein the creating of the goal includes selecting a value from the last of the variable value layers and randomly selecting at least some of the plurality of variables.
2. The method of claim 1 , wherein the generating of the plurality of atoms includes randomly selecting a number of the plurality of atoms, and wherein the partitioning the plurality of atoms into the plurality of variables includes randomly selecting a number of the plurality of variables.
3. The method of claim 1 , wherein the generated planning task is a simplified action structures (SAS+) planning task, and further comprising translating the SAS+ planning task to a Stanford Research Institute Problem Solver (STRIPS) fragment of Planning Domain Definition Language (PDDL).
4. The method of claim 1 , wherein the casual graph includes a structure, and the structure includes a chain, a bipartite graph, a fork, a tree, or a direct acyclic graph (DAG).
5. The method of claim 1 , wherein the layered graph is a relaxed planning graph (RPG).
6. A system for creating a planning task comprising:
a processor executing instructions stored in a memory device, wherein the processor:
generates a plurality of atoms;
partitions the plurality of atoms into a plurality of variables;
generates a casual graph based on the plurality of variables;
creates a layered graph including interchanging variable value layers and action layers based on the casual graph, wherein the creating of the layered graph is ceased when the last of the variable value layers includes all of the plurality of atoms and the casual graph is fully covered; and
generates a planning task based on the layered graph according to a goal created for the planning task, wherein the creating of the goal includes selecting a value from the last of the variable value layers and randomly selecting at least some of the plurality of variables.
7. The system of claim 6 , wherein the generating of the plurality of atoms includes randomly selecting a number of the plurality of atoms, and wherein the partitioning the plurality of atoms into the plurality of variables includes randomly selecting a number of the plurality of variables.
8. The system of claim 6 , wherein the generated planning task is a simplified action structures (SAS+) planning task, and wherein the processor further translates the SAS+ planning task to a Stanford Research Institute Problem Solver (STRIPS) fragment of Planning Domain Definition Language (PDDL).
9. The system of claim 6 , wherein the casual graph includes a structure, and the structure includes a chain, a bipartite graph, a fork, a tree, or a direct acyclic graph (DAG).
10. The system of claim 6 , wherein the layered graph is a relaxed planning graph (RPG).
11. A computer program product for creating a planning task, by a processor, the computer program product embodied on a non-transitory computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions comprising:
an executable portion that generates a plurality of atoms;
an executable portion that partitions the plurality of atoms into a plurality of variables;
an executable portion that generates a casual graph based on the plurality of variables;
an executable portion that creates a layered graph including interchanging variable value layers and action layers based on the casual graph, wherein the creating of the layered graph is ceased when the last of the variable value layers includes all of the plurality of atoms and the casual graph is fully covered; and
an executable portion that generates a planning task based on the layered graph according to a goal created for the planning task, wherein the creating of the goal includes selecting a value from the last of the variable value layers and randomly selecting at least some of the plurality of variables.
12. The computer program product of claim 11 , wherein the generating of the plurality of atoms includes randomly selecting a number of the plurality of atoms, and wherein the partitioning the plurality of atoms into the plurality of variables includes randomly selecting a number of the plurality of variables.
13. The computer program product of claim 11 , wherein the generated planning task is a simplified action structures (SAS+) planning task, and wherein the computer-readable program code portions further include an executable portion that translates the SAS+ planning task to a Stanford Research Institute Problem Solver (STRIPS) fragment of Planning Domain Definition Language (PDDL).
14. The computer program product of claim 11 , wherein the casual graph includes a structure, and the structure includes a chain, a bipartite graph, a fork, a tree, or a direct acyclic graph (DAG).
15. The computer program product of claim 11 , wherein the layered graph is a relaxed planning graph (RPG).