Policy information determining method and device

Abstract

The invention discloses a policy information determining method. The method comprises the steps of obtaining data to be processed, wherein the data to be processed comprises a first image and a secondimage, the first image includes scene thumbnail data of an interactive application, and the second image comprises scene interactive data of the interactive application; processing the first image through a policy information model to obtain target action data, wherein the policy information model is the relation between label sample data and sample action data; according to the second image, determining target state data of a target object; according to the target action data and the target state data, determining policy information corresponding to the target object, wherein the policy information is used for controlling target object actions. The invention further discloses a policy information determining device. Game operation automation can be achieved, for the complex game scene, the policy information obtained by means of the policy information model is closer to actions of a real player, and therefore the flexibility of the scheme is improved.

Description

technical field

[0001] The invention relates to the field of artificial intelligence, in particular to a method and device for determining policy information. Background technique

[0002] At present, interactive applications usually have a mode of human-computer interaction, or a mode of machine simulating user operations. For example, in interactive applications, behavior trees can be used to judge and output action strategies. When the terminal device judges the action policy, it needs to search the behavior tree from top to bottom through some conditions, and finally determine the policy information to be executed.

[0003] However, although policy information can be output by judging the specified conditions in the behavior tree, since the specified conditions in the behavior tree are limited, it is not very good for complex interactive application scenarios without specifying Therefore, the flexibility of using behavior trees is low. Contents of the invention [00...

Images