Systems and methods for using natural language processing (NLP) to control automated execution of in-game activities

Patent metadata

Field	Value
Publication number	US-12090397-B2
Application number	US-202117315817-A
Country	US
Kind code	B2
Filing date	May 10, 2021
Priority date	Oct 9, 2020
Publication date	Sep 17, 2024
Grant date	Sep 17, 2024

Abstract

Systems and methods for using natural language processing (NLP) to control automated gameplay in one or more online games within an online gaming platform are disclosed. Exemplary implementations may: train a pretrained text-based system to generate sequences indicating one or more activities, wherein activities are associated with characteristics of user-controllable characters such that a particular user-controllable character having a particular characteristic may perform, within the online gaming platform, one or more activities from a particular set that is associated with the particular characteristic; receive user input describing one or more commands; generate, using the pretrained text-based system, a sequence indicating one or more activities based on the user input; transfer the generated sequence to the online gaming platform; effectuate automated execution of activities by the particular user-controllable character in accordance with the generated sequence.

First claim

What is claimed is:

1. A system configured to use natural language processing (NLP) to control automated gameplay in an online game within an online gaming platform, the system comprising:
electronic storage configured to electronically store information, wherein the information includes a set of in-game activities in the online game, wherein at least some of the in-game activities are associated with a particular activity-defining characteristic of a user-controllable character, wherein the particular activity-defining characteristic defines particular in-game activities that are available to be performed during the automated gameplay;
one or more hardware processors configured by machine-readable instructions to:
determine, based on the stored information, the particular in-game activities that are associated with the particular activity-defining characteristic, wherein the particular in-game activities are available to be performed during the automated gameplay by the user-controllable character;
train a pretrained text-based system, using prompts, to generate sequences including one or more identifiers of the particular in-game activities, wherein individual ones of the prompts are based on a combination of:
(i) a textual description of an individual in-game activity, and
(ii) an identifier of the individual in-game activity;

receive, from a user, a user input, wherein the user input represents a user-provided textual description describing multiple commands for the user-controllable character, wherein the user-provided textual description further describes a condition to be evaluated by the user-controllable character during the automated gameplay by the user-controllable character, wherein at least one of the multiple commands is controlled by the condition;
generate, using the trained pretrained text-based system, a sequence including the condition and at least a first identifier of a first in-game activity and a second identifier of a second in-game activity, wherein generation of the generated sequence is based on the user-provided textual description;
transfer the generated sequence to the online gaming platform; and
effectuate automated execution of in-game activities by the user-controllable character in accordance with the generated sequence, wherein the automated execution includes an evaluation of the condition, wherein, subsequent to and depending on the evaluation of the condition by the user-controllable character during the automated gameplay by the user-controllable character, either the first in-game activity or the second in-game activity is automatically executed during the automated gameplay.

2. The system of claim 1 , wherein at least part of the online game takes place in a virtual space, and wherein the evaluation of the condition is independent and separate from the first in-game activity and from the second in-game activity.

3. The system of claim 1 , wherein the multiple commands include one or more commands directed to a given user-controllable character, wherein the one or more hardware processors are further configured to:
determine a given activity-defining characteristic for the given user-controllable character, wherein the given activity-defining characteristic is different from the particular activity-defining characteristic;
determine, based on the stored information, an appropriate characteristic-specific set of in-game activities for the given user-controllable character; and
generate, using the trained pretrained text-based system, a given sequence of given in-game activities selected from the appropriate characteristic-specific set of in-game activities;
wherein the given sequence is different from the generated sequence.

4. The system of claim 1 , wherein a first individual activity-defining characteristic of a first user-controllable character is associated with a first characteristic-specific set of in-game activities, wherein a second individual activity-defining characteristic of a second user-controllable character is associated with a second characteristic-specific set of in-game activities, and wherein the first characteristic-specific set of in-game activities is different from the second characteristic-specific set of in-game activities.

5. The system of claim 1 , wherein the set of in-game activities is specific to an individual online game within the online gaming platform.

6. The system of claim 1 , wherein the particular in-game activities correspond to an individual race and/or species of the user-controllable character.

7. The system of claim 1 , wherein the one or more hardware processors are further configured to:
select the user-controllable character from multiple user-controllable characters; and
determine the particular activity-defining characteristic for the user-controllable character.

8. The system of claim 3 , wherein the user-provided textual description corresponds to a first particular in-game activity responsive to the given user-controllable character having a first activity-defining characteristic, and wherein the user-provided textual description corresponds to a second particular in-game activity responsive to the given user-controllable character having a second activity-defining characteristic that is different from the first activity-defining characteristic, wherein the first particular in-game activity is different from the second particular in-game activity.

9. The system of claim 1 , wherein the pretrained text-based system is Generative Pretrained Transformer 3 (GPT-3).

10. The system of claim 1 , wherein the user input is received through a user interface.

11. A method to use natural language processing (NLP) to control automated gameplay in one or more online games within an online gaming platform, the method being implemented in a computer system, the method comprising:
electronically storing information, wherein the information includes a set of in-game activities in the online game, wherein at least some of the in-game activities are associated with a particular activity-defining characteristic of a user-controllable character, wherein the particular activity-defining characteristic defines particular in-game activities that are available to be performed during the automated gameplay;
determining, based on the stored information, the particular in-game activities that are associated with the particular activity-defining characteristic, wherein the particular in-game activities are available to be performed during the automated gameplay by the user-controllable character;
training a pretrained text-based system, using prompts, to generate sequences including one or more identifiers of the particular in-game activities, wherein individual ones of the prompts are based on a combination of:
(i) a textual description of an individual in-game activity, and
(ii) an identifier of the individual in-game activity;

receiving, from a user, a user input, wherein the user input represents a user-provided textual description describing multiple commands for the user-controllable character, wherein the user-provided textual description further describes a condition to be evaluated by the user-controllable character during the automated gameplay by the user-controllable character, wherein at least one of the multiple commands is controlled by the condition;
generating, using the trained pretrained text-based system, a sequence including the condition and at least a first identifier of a first in-game activity and a second identifier of a second in-game activity, wherein generation of the generated sequence is based on the user-provided textual description;
transferring the generated sequence to the online

Assignees

• Mythical Inc

Inventors

• Thacker Cameron
• Linden John

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• A63F13/215

  comprising means for detecting acoustic signals, e.g. using a microphone · CPC title

• A63F2300/1081

  Input via voice recognition · CPC title

• G06N20/00

  Machine learning · CPC title