Granted Patent B2

US 12,572,819 · App. 17/805,572 · Granted Mar 10, 2026

Inventors: Michael Katz (Goldens Bridge, NY); Shirin Sohrabi Araghi (Briarcliff Manor, NY)

Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATION

G06N3/126

Abstract

Claims (60)

1 . A computer-implemented method comprising:

formulating, by one or more processors, an artificial intelligence (AI) planning problem represented as a computer-readable model comprising a set of operators;

executing a planning algorithm, by the one or more processors, to generate a solution set to the AI planning problem comprising one or more plans;

iteratively reformulating, by the one or more processors, the computer-readable model of the AI planning problem by encoding exclusion conditions such that the planning algorithm cannot generate:

a previously generated plan from a previous iteration; and

any plan that, for each operator occurring a given number of times in the previously generated plan, includes that operator at least that given number of times; and

executing the planning algorithm, by the one or more processors, on the iteratively reformulated computer-readable model to generate additional plans until a pre-defined stopping criterion is reached, wherein executing the planning algorithm on the iteratively reformulated computer-readable model reduces a size of the solution set, thereby improving computational efficiency.

2 . The computer-implemented method of claim 1 , wherein the encoding exclusion conditions further comprises:

extending to track one or more operators on the previously generated plan;

modifying and extending the one or more operators to change a value of an extra variable;

extending an initial state to allocate an initial value to the extra variable; and

extending a goal to allocate a goal value to the one or more operators and to check the extra variable.

3 . The computer-implemented method of claim 1 , wherein the iteratively reformulating the computer-readable model of the AI planning problem further comprises:

forbidding plans that correspond to super-multisets of the one or more plans.

4 . The computer-implemented method of claim 1 , wherein the pre-defined stopping criterion is selected from the group consisting of: a plan of worse quality than a quality bound is found, enough plans are found, and no more plans exist.

5 . The computer-implemented method of claim 1 , wherein the planning algorithm is selected from the group consisting of: a cost-optimal planner, an agile planner, a satisficing planner, a top-k planner, a top-quality planner, and a diverse planner.

6 . The computer-implemented method of claim 1 , further comprising:

generating a subset top-quality solution in domain with infinite size unordered top-quality solutions.

7 . The computer-implemented method of claim 1 , further comprising:

utilizing a cost-optimal planner to generate an updated solution to the AI planning problem based on the pre-defined stopping criterion.

8 . A computer program product comprising:

one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions comprising:

program instructions to formulate an artificial intelligence (AI) planning problem represented as a computer-readable model comprising a set of operators;

program instructions to execute a planning algorithm to generate a solution set to the AI planning problem comprising one or more plans;

program instructions to iteratively reformulate the computer-readable model of the AI planning problem by encoding exclusion conditions such that the planning algorithm cannot generate:

a previously generated plan from a previous iteration; and

any plan that, for each operator occurring a given number of times in the previously generated plan, includes that operator at least that given number of times; and

program instructions to execute the planning algorithm on the iteratively reformulated computer-readable model to generate additional plans until a pre-defined stopping criterion is reached, wherein executing the planning algorithm on the iteratively reformulated computer-readable model reduces a size of the solution set, thereby improving computational efficiency.

9 . The computer program product of claim 8 , wherein the encoding exclusion conditions further comprises:

program instructions to extend to track one or more operators on the previously generated plan;

program instructions to modify and extend the one or more operators to change a value of an extra variable;

program instructions to extend an initial state to allocate an initial value to the extra variable; and

program instructions to extend a goal to allocate a goal value to the one or more operators and to check the extra variable.

10 . The computer program product of claim 8 , wherein the program instructions to iteratively reformulate the computer-readable model of the AI planning problem further comprises:

program instructions to forbid plans that correspond to super-multisets of the one or more plans.

11 . The computer program product of claim 8 , wherein the pre-defined stopping criterion is selected from the group consisting of: a plan of worse quality than a quality bound is found, enough plans are found, and no more plans exist.

12 . The computer program product of claim 8 , wherein the planning algorithm is selected from the group consisting of: a cost-optimal planner, an agile planner, a satisficing planner, a top-k planner, a top-quality planner, and a diverse planner.

13 . The computer program product of claim 8 , further comprising:

program instructions to generate a subset top-quality solution in domain with infinite size unordered top-quality solutions.

14 . The computer program product of claim 8 , further comprising:

program instructions to utilize a cost-optimal planner to generate an updated solution to the AI planning problem based on the pre-defined stopping criterion.

15 . A computer system comprising:

one or more computer processors, one or more computer readable storage media, and program instructions stored on the one or more computer readable storage media for execution by at least one of the one or more computer processors, the program instructions comprising:

program instructions to formulate an artificial intelligence (AI) planning problem represented as a computer-readable model comprising a set of operators;

program instructions to execute a planning algorithm to generate a solution set to the AI planning problem comprising one or more plans;

program instructions to iteratively reformulate the computer-readable model of the AI planning problem by encoding exclusion conditions such that the planning algorithm cannot generate:

a previously generated plan from a previous iteration; and

any plan that, for each operator occurring a given number of times in the previously generated plan, includes that operator at least that given number of times; and

program instructions to execute the planning algorithm on the iteratively reformulated computer-readable model to generate additional plans until a pre-defined stopping criterion is reached, wherein executing the planning algorithm on the iteratively reformulated computer-readable model reduces a size of the solution set, thereby improving computational efficiency.

16 . The computer system of claim 15 , wherein the encoding exclusion conditions further comprises:

program instructions to extend to track one or more operators on the previously generated plan;

program instructions to modify and extend the one or more operators to change a value of an extra variable;

program instructions to extend an initial state to allocate an initial value to the extra variable; and

program instructions to extend a goal to allocate a goal value to the one or more operators and to check the extra variable.

17 . The computer system of claim 15 , wherein the program instructions to iteratively reformulate the computer-readable model of the AI planning problem further comprises:

program instructions to forbid plans that correspond to super-multisets of the one or more plans.

18 . The computer system of claim 15 , wherein the pre-defined stopping criterion is selected from the group consisting of: a plan of worse quality than a quality bound is found, enough plans are found, and no more plans exist.

19 . The computer system of claim 15 , wherein the planning algorithm is selected from the group consisting of: a cost-optimal planner, an agile planner, a satisficing planner, a top-k planner, a top-quality planner, and a diverse planner.

20 . The computer system of claim 15 , further comprising:

program instructions to generate a subset top-quality solution in domain with infinite size unordered top-quality solutions.

Assignments (1)

Continuity (1)

References Cited (31)