Multi-agent plan recognition

1. A computer-implemented method, comprising:

transforming, by a device operatively coupled to a processor, a problem model and an at least partially ordered sequence of observations into an artificial intelligence planning problem through a transform algorithm, wherein the problem model comprises a domain description from a plurality of agents and a durative action, wherein at least one of the observations of the at least partially ordered sequence of observations is a condition that changes over time, and wherein the plurality of agents comprises an agent operated in accordance with a plan;

determining, by the device, plan information corresponding to the plan using an artificial intelligence planner on the artificial intelligence planning problem, wherein the plan information is determined by the artificial intelligence planner by determining a posterior probability of the agent operated in accordance with the plan, wherein the artificial intelligence planner determines the posterior probability two times per goal, wherein the determining the posterior probability employs an approach that comprises assigning a cost difference between a first cost of an optimal plan to achieve a goal comprising one or more observations and a second cost of the optimal plan to achieve the goal prior to the one or more observations, and wherein a combined difference between the first cost of the optimal plan and the second cost of the optimal plan determines the posterior probability, based on the at least partially ordered sequence of the one or more observations of one or more prior occurrences; and

translating, by the device, the plan information into information indicative of a solution to the artificial intelligence planning problem.

2. The computer-implemented method of claim 1 , wherein the partially ordered sequence of observations comprises an action condition, and wherein the computer-implemented method further comprises:

translating, by the device, the action condition into a second condition that changes over time.

3. The computer-implemented method of claim 1 , wherein a description of the problem model comprises an attribute selected from the group consisting of one or more facts, an initial state, one or more action operators, a goal condition, and a number of agents of the plurality of agents in the problem model.

4. The computer-implemented method of claim 1 , wherein a description of the problem model comprises a defined initial state or a defined goal condition.

5. The computer-implemented method of claim 1 , wherein the determining the plan information comprises determining the plan information employing a top-K planning component.

6. The computer-implemented method of claim 1 , wherein the determining the plan information comprises determining the plan information employing a diverse planning component.

7. The computer-implemented method of claim 1 , wherein the determining the plan information comprises determining the plan information employing a delta planning component.

8. The computer-implemented method of claim 1 , wherein the determining the plan information comprises determining the plan information employing a hybrid planning component.

9. The computer-implemented method of claim 1 , wherein the at least partially ordered sequence of observations comprises an unreliable observation, and wherein the unreliable observation is an observation selected from the group consisting of an inconsistent observation, a noisy observation, and a missing observation.

10. The computer-implemented method of claim 1 , wherein the transforming comprises maintaining, in the artificial intelligence planning problem, one or more prior access restrictions of the plurality of agents relative to one another.

11. The computer-implemented method of claim 1 , wherein the agent operated in accordance with the plan is an agent operated in accordance with the plan to achieve the goal, and wherein the computer-implemented method further comprises:

determining, by the device, goal information corresponding to the goal using the artificial intelligence planner on the artificial intelligence planning problem, by determining a posterior probability of the agent operated to achieve the goal, based on the at least partially ordered sequence of observations.

12. The computer-implemented method of claim 1 , wherein the transforming comprises preserving, by the device, the at least partially ordered sequence of the observations in the artificial intelligence planning problem.

13. The computer-implemented method of claim 1 , wherein the determining comprises applying, by the device, a penalty relative to discarding by the artificial intelligence planner of an observation of the at least partially ordered sequence of the observations.

14. A computer program product facilitating a process to infer a solution to a multi-agent problem model, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:

transform, by the processor, a problem model and an at least partially ordered sequence of observations into an artificial intelligence planning problem through a transform algorithm, wherein the problem model comprises a domain description from a plurality of agents and a durative action, wherein at least one of the observations of the at least partially ordered sequence of observations is a condition that changes over time, and wherein the plurality of agents comprises an agent operated in accordance with a plan;

determine, by the processor, plan information corresponding to the plan using an artificial intelligence planner on the artificial intelligence planning problem, wherein the plan information is determined by the artificial intelligence planner by determining a posterior probability of the agent operated in accordance with the plan, wherein the artificial intelligence planner determines the posterior probability two times per goal, wherein the determining the posterior probability employs an approach that comprises assigning a cost difference between a first cost of an optimal plan to achieve a goal comprising one or more observations and a second cost of the optimal plan to achieve the goal prior to the one or more observations, and wherein a combined difference between the first cost of the optimal plan and the second cost of the optimal plan determines the posterior probability, based on the at least partially ordered sequence of the one or more observations of one or more prior occurrences; and

translate, by the processor, the plan information into information indicative of a solution to the artificial intelligence planning problem.

15. The computer program product of claim 14 , wherein the program instructions are further executable by the processor to cause the processor to:

maintain, by the processor, in the artificial intelligence planning problem relative to the transforming, one or more prior access restrictions of the plurality of agents relative to one another.

16. The computer program product of claim 14 , wherein the agent operated in accordance with the plan is an agent operated in accordance with the plan to achieve a goal, and wherein the program instructions are further executable by the processor to cause the processor to:

determining, by the processor, goal information corresponding to the goal using the artificial intelligence planner on the artificial intelligence planning problem, including determining, by the processor, a posterior probability of the agent operated to achieve the goal, based on the at least partially ordered sequence of observations.

17. The computer program product of claim 14 , wherein the program instructions are further executable by the processor to cause the processor to:

preserve, by the processor, relative to the transforming, the at least partially ordered sequence of the observations in the artificial intelligence planning problem.

18. The computer program product of claim 14 , wherein the program instructions are further executable by the processor to cause the processor to:

apply, by the processor, relative to the determining, a penalty relative to discarding by the artificial intelligence planner of an observation of the at least partially ordered sequence of the observations.

19. The computer program product of claim 14 , wherein the partially ordered sequence of observations comprises an action condition, and wherein the program instructions are further executable by the processor to cause the processor to:

translate, by the processor, the action condition into a second condition that changes over time.

20. The computer program product of claim 14 , wherein the at least partially ordered sequence of observations comprises an unreliable observation, and wherein the unreliable observation is an observation selected from the group consisting of an inconsistent observation, a noisy observation, and a missing observation.