Method and device for testing artificial intelligence behavior in game and electronic equipment

Abstract

The invention relates to a method and a device for testing an artificial intelligence behavior in a game and electronic equipment. The method comprises the following steps: modeling an artificial intelligence behavior tree in a configuration file to obtain a tree object; performing node splitting on the tree object to obtain a node set, performing branch splitting on the tree object to obtain a branch set and respectively recording execution conditions of branches in the branch set; respectively loading nodes in the node set into the game for test, respectively loading the branches in the branch set into the game for test after the recoded execution conditions are set and respectively loading the tree object into the game for test after the recoded execution conditions are set. Through the method and the device provided by the invention, the test efficiency aiming at an AI behavior in the game test can be improved.

Description

technical field [0001] The present application relates to the technical field of software testing, in particular, to a method, device and electronic equipment for testing artificial intelligence behavior in games. Background technique [0002] In the current major games, whether it is a terminal game or a mobile game, there are more or less game characters such as NPCs and monsters in the dungeon, and these game characters have a set of fixed logic to control them. They simulate real character behaviors, and these game behaviors are usually called AI behaviors. As small as a mob in a dungeon, as large as a big boss in a dungeon, they all rely on AI behaviors of different complexity to control. In a game, with the increase of monsters and bosses, the number of AI behaviors also increases. A large game may contain hundreds or even thousands of different AI behavior logics. In game development, an AI behavior is controlled by a text based on XML format, which is similar to th...

AI Technical Summary

Problems solved by technology

First of all, when it comes to numerical correlation, the results cannot be obtained accurately, and it can only be estimated manually, which is inaccurate and inefficient

Secondly, some AI behavior tree branches may be missed through self-combination, resulting in incomplete test coverage

In addition, the test logic of this method is not sound. It does not start from the most basic single behavior tree combination node and the basic function node of the game. If there is a behavior tree branch with a single node with defective functions later, the execution of this branch will fail. It may mislead testers to judge that it is a problem with other nodes and conditions of this branch

Moreover, this method lacks the relevant parts of result collection, and in the face of a large number of AI behavior trees, it cannot intuitively give the execution status of each tree

Images