Action control method and apparatus

1. An action control method comprising:

obtaining states of N dimensions of an artificial intelligence device, wherein N is a positive integer greater than or equal to one;

obtaining a plurality of discrete decisions, based on an active fuzzy subset and a control model, corresponding to a state of each of the N dimensions, wherein the active fuzzy subset is a fuzzy subset comprises a membership degree of the state that is not zero, wherein the fuzzy subset is a state interval that corresponds to a same discrete decision in a dimension, wherein the membership degree indicates a degree that the state belongs to the fuzzy subset, and wherein the control model outputs a corresponding discrete decision based on an input state;

performing, based on a membership degree between a state and an active fuzzy subset that are of each dimension, a weighted summation on the discrete decisions to obtain a continuous decision; and

controlling, based on the continuous decision, the artificial intelligence device to execute a corresponding action.

2. The action control method of claim 1 , further comprising:

obtaining, for each of the discrete decisions, membership degrees of N active fuzzy subsets corresponding to each discrete decision to obtain N membership degrees;

calculating, based on the N membership degrees, weights of the discrete decisions; and

performing, based on the weights of the discrete decisions, the weighted summation on the discrete decisions to obtain the continuous decision.

3. The action control method of claim 1 , further comprising:

obtaining a central value of an active fuzzy subset of each of the N dimensions to obtain a plurality of central values;

combining central values of the N dimensions to obtain a plurality of intermediate states; and

inputting the intermediate states into the control model to obtain the discrete decisions from the control model.

4. The action control method of claim 1 , wherein before obtaining the discrete decisions, the action control method further comprises setting, for each of the N dimensions, each fuzzy subset as an active fuzzy subset of each dimension when the membership degree is not zero.

5. The action control method of claim 1 , wherein before obtaining the states of the N dimensions of the artificial intelligence device, the action control method further comprises:

dividing, for each of the N dimensions, a state space of each dimension into a plurality of state intervals;

obtaining, based on the control model, a typical discrete decision of each of the state intervals to obtain a plurality of typical discrete decisions; and

combining, based on the typical discrete decisions, a plurality of adjacent state intervals corresponding to a same typical discrete decision into a fuzzy subset to obtain a fuzzy subset of each dimension.

6. The action control method of claim 5 , further comprising:

obtaining, for each of the state intervals, a plurality of representative states of each state interval, wherein each representative state comprises a central value of each state interval of each dimension and any one of other states of each dimension;

respectively inputting the representative states into the control model to obtain the discrete decisions from the control model; and

selecting, from the discrete decisions, a discrete decision with a maximum quantity of repetitions as a typical discrete decision of each state interval.

7. The action control method of claim 1 , wherein after obtaining the N dimensions of the artificial intelligence device, the action control method further comprises calculating, for each fuzzy subset of each of the N dimensions, a state of each dimension using a membership function corresponding to each fuzzy subset to obtain a membership degree of each fuzzy subset.

8. The action control method of claim 7 , wherein before obtaining the states of the N dimensions of the artificial intelligence device, the action control method further comprises obtaining, based on a preset rule, the membership function of each fuzzy subset, wherein the membership function calculates a membership degree of a corresponding fuzzy subset, and wherein the preset rule is that a central value of the membership function in each fuzzy subset is set to one, a membership degree of a boundary value in each fuzzy subset is set to 0.5, and a central value of two adjacent fuzzy subsets in all the fuzzy subsets is set to zero.

9. The action control method of claim 1 , wherein before obtaining the discrete decisions, the action control method further comprises selecting, for each of the N dimensions, two fuzzy subsets as active fuzzy subsets of each dimension from a plurality of fuzzy subsets of each dimension, and wherein central values of the two fuzzy subsets are around the state of each of the N dimensions.

10. An action control device comprising:

a non-transitory medium configured to store program instructions; and

a processor coupled to the a non-transitory medium, wherein the program instructions cause the processor to be configured to:

obtain states of N dimensions of an artificial intelligence device, wherein N is a positive integer greater than or equal to one;

obtain a plurality of discrete decisions based on an active fuzzy subset and a control model corresponding to a state of each of the N dimensions, wherein the active fuzzy subset is a fuzzy subset comprises a membership degree of the state that is not zero, wherein the fuzzy subset comprises a state interval that corresponds to a same discrete decision in a dimension, wherein the membership degree indicates a degree that the state belongs to the fuzzy subset, and wherein the control model outputs a corresponding discrete decision based on an input state;

perform, based on a membership degree between a state and an active fuzzy subset that are of each dimension, a weighted summation on the discrete decisions, to obtain a continuous decision; and

control, based on the continuous decision, the artificial intelligence device to execute a corresponding action.

11. The action control device of claim 10 , wherein the program instructions further cause the processor to be configured to:

obtain, for each of the discrete decisions, membership degrees of N active fuzzy subsets corresponding to each discrete decision to obtain N membership degrees;

calculate, based on the N membership degrees, weights of the discrete decisions; and

perform, based on the weights of the discrete decisions, the weighted summation on the discrete decisions to obtain the continuous decision.

12. The action control device of claim 10 , wherein the program instructions further cause the processor to be configured to:

obtain a central value of an active fuzzy subset of each of the N dimensions to obtain a plurality of central values;

combine central values of the N dimensions to obtain a plurality of intermediate states; and

respectively input the intermediate states into the control model to obtain the plurality of discrete decisions from the control model.

13. The action control device of claim 10 , wherein the program instructions further cause the processor to be configured to set, for each of the N dimensions, each fuzzy subset as an active fuzzy subset of each dimension when a membership degree between a state and any fuzzy subset that are of each dimension is not zero.

14. The action control device of claim 10 , wherein the program instructions further cause the processor to be configured to:

divide, for each of the N dimensions, a state space of each dimension into a plurality of state intervals;

obtain, based on the control model, a typical discrete decision of each of the plurality of state intervals to obtain a plurality of typical discrete decisions; and

combine, based on the typical discrete decisions, a plurality of adjacent state intervals corresponding to a same typical discrete decision into a fuzzy subset to obtain a fuzzy subset of each dimension.

15. The action control device of claim 14 , wherein the program instructions further cause the processor to be configured to:

obtain, for each of the state intervals, a plurality of representative states of each state interval, wherein each representative state comprises a central value of each state interval of each dimension and any one of other states of each dimension;

respectively input the representative states into the control model to obtain the discrete decisions from the control model; and

select, from the discrete decisions, a discrete decision with a maximum quantity of repetitions as a typical discrete decision of each state interval.

16. The action control device of claim 10 , wherein the program instructions further cause the processor to be configured to calculate, for each fuzzy subset of each of the N dimensions, a state of each dimension using a membership function corresponding to each fuzzy subset to obtain a membership degree of each fuzzy subset.

17. The action control device of claim 16 , wherein the program instructions further cause the processor to be configured to obtain, based on a preset rule, the membership function of each fuzzy subset, wherein the membership function calculates a membership degree of a fuzzy subset of a corresponding dimension, and wherein the preset rule is that a central value of the membership function in each fuzzy subset is set to one, a membership degree of a boundary value in each fuzzy subset is set to 0.5, and a central value of two adjacent fuzzy subsets in all the fuzzy subsets is set to zero.

18. The action control device of claim 10 , wherein the program instructions further cause the processor to be configured to select, for each of the N dimensions, two fuzzy subsets as active fuzzy subsets of each dimension from a plurality of fuzzy subsets of each dimension, and wherein central values of the two fuzzy subsets are around the state of each of the N dimensions.

19. A computer program product comprising computer-executable instructions for storage on a non-transitory computer-readable storage medium that, when executed by a processor, cause an apparatus to:

obtain states of N dimensions of an artificial intelligence device, wherein N is a positive integer greater than or equal to one;

obtain a plurality of discrete decisions based on an active fuzzy subset and a control model corresponding to a state of each of the N dimensions, wherein the active fuzzy subset is a fuzzy subset comprises a membership degree of the state that is not zero, wherein the fuzzy subset is a state interval that corresponds to a same discrete decision in a dimension, wherein the membership degree indicates a degree that the state belongs to the fuzzy subset, and wherein the control model outputs a corresponding discrete decision based on an input state;

perform, based on a membership degree between a state and an active fuzzy subset that are of each dimension, a weighted summation on the discrete decisions to obtain a continuous decision; and

control, based on the continuous decision, the artificial intelligence device to execute a corresponding action.

20. The computer program product of claim 19 , wherein the computer-executable instructions further cause the apparatus to:

obtain, for each of the discrete decisions, membership degrees of N active fuzzy subsets corresponding to each discrete decision to obtain N membership degrees;

calculate, based on the N membership degrees, weights of the discrete decisions; and

perform, based on the weights of the discrete decisions, the weighted summation on the discrete decisions to obtain the continuous decision.